idnits 2.17.1 draft-ietf-cdni-framework-09.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (January 30, 2014) is 3737 days in the past. Is this intentional? -- Found something which looks like a code comment -- if you have code sections in the document, please surround them with '' and '' lines. Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-15) exists of draft-ietf-cdni-control-triggers-02 == Outdated reference: A later version (-20) exists of draft-ietf-cdni-footprint-capabilities-semantics-01 == Outdated reference: A later version (-27) exists of draft-ietf-cdni-logging-08 == Outdated reference: A later version (-21) exists of draft-ietf-cdni-metadata-04 == Outdated reference: A later version (-20) exists of draft-ietf-cdni-redirection-01 == Outdated reference: A later version (-17) exists of draft-ietf-cdni-requirements-16 -- Obsolete informational reference (is this intentional?): RFC 3466 (Obsoleted by RFC 7336) Summary: 0 errors (**), 0 flaws (~~), 7 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group L. Peterson 3 Internet-Draft Akamai Technologies, Inc. 4 Obsoletes: 3466 (if approved) B. Davie 5 Intended status: Informational VMware, Inc. 6 Expires: August 3, 2014 R. van Brandenburg, Ed. 7 TNO 8 January 30, 2014 10 Framework for CDN Interconnection 11 draft-ietf-cdni-framework-09 13 Abstract 15 This document presents a framework for Content Distribution Network 16 Interconnection (CDNI). The purpose of the framework is to provide 17 an overall picture of the problem space of CDNI and to describe the 18 relationships among the various components necessary to interconnect 19 CDNs. CDN Interconnection requires the specification of interfaces 20 and mechanisms to address issues such as request routing, 21 distribution metadata exchange, and logging information exchange 22 across CDNs. The intent of this document is to outline what each 23 interface needs to accomplish, and to describe how these interfaces 24 and mechanisms fit together, while leaving their detailed 25 specification to other documents. It obsoletes RFC 3466. 27 Status of This Memo 29 This Internet-Draft is submitted in full conformance with the 30 provisions of BCP 78 and BCP 79. 32 Internet-Drafts are working documents of the Internet Engineering 33 Task Force (IETF). Note that other groups may also distribute 34 working documents as Internet-Drafts. The list of current Internet- 35 Drafts is at http://datatracker.ietf.org/drafts/current/. 37 Internet-Drafts are draft documents valid for a maximum of six months 38 and may be updated, replaced, or obsoleted by other documents at any 39 time. It is inappropriate to use Internet-Drafts as reference 40 material or to cite them other than as "work in progress." 42 This Internet-Draft will expire on August 3, 2014. 44 Copyright Notice 46 Copyright (c) 2014 IETF Trust and the persons identified as the 47 document authors. All rights reserved. 49 This document is subject to BCP 78 and the IETF Trust's Legal 50 Provisions Relating to IETF Documents 51 (http://trustee.ietf.org/license-info) in effect on the date of 52 publication of this document. Please review these documents 53 carefully, as they describe your rights and restrictions with respect 54 to this document. Code Components extracted from this document must 55 include Simplified BSD License text as described in Section 4.e of 56 the Trust Legal Provisions and are provided without warranty as 57 described in the Simplified BSD License. 59 Table of Contents 61 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 62 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . . . 3 63 1.2. Reference Model . . . . . . . . . . . . . . . . . . . . . 4 64 1.3. Structure Of This Document . . . . . . . . . . . . . . . 8 65 2. Building Blocks . . . . . . . . . . . . . . . . . . . . . . . 8 66 2.1. Request Redirection . . . . . . . . . . . . . . . . . . . 8 67 2.1.1. DNS Redirection . . . . . . . . . . . . . . . . . . . 8 68 2.1.2. HTTP Redirection . . . . . . . . . . . . . . . . . . 9 69 3. Overview of CDNI Operation . . . . . . . . . . . . . . . . . 10 70 3.1. Preliminaries . . . . . . . . . . . . . . . . . . . . . . 12 71 3.2. Iterative HTTP Redirect Example . . . . . . . . . . . . . 13 72 3.3. Recursive HTTP Redirection Example . . . . . . . . . . . 18 73 3.4. Iterative DNS-based Redirection Example . . . . . . 22 74 3.5. Dynamic Footprint Discovery Example . . . . . . . . . . . 25 75 3.6. Content Removal Example . . . . . . . . . . . . . . . . . 27 76 3.7. Pre-Positioned Content Acquisition Example . . . . . . . 28 77 3.8. Asynchronous CDNI Metadata Example . . . . . . . . . . . 29 78 3.9. Synchronous CDNI Metadata Acquisition Example . . . . . . 31 79 3.10. Content and Metadata Acquisition with Multiple Upstream 80 CDNs . . . . . . . . . . . . . . . . . . . . . . . . . . 33 81 4. Main Interfaces . . . . . . . . . . . . . . . . . . . . . . . 34 82 4.1. In-Band versus Out-of-Band Interfaces . . . . . . . . . . 35 83 4.2. Cross Interface Concerns . . . . . . . . . . . . . . . . 35 84 4.3. Request Routing Interfaces . . . . . . . . . . . . . . . 36 85 4.4. CDNI Logging Interface . . . . . . . . . . . . . . . . . 37 86 4.5. CDNI Control Interface . . . . . . . . . . . . . . . . . 39 87 4.6. CDNI Metadata Interface . . . . . . . . . . . . . . . . . 39 88 4.7. HTTP Adaptive Streaming Concerns . . . . . . . . . . . . 40 89 4.8. URI Rewriting . . . . . . . . . . . . . . . . . . . . . . 42 90 5. Deployment Models . . . . . . . . . . . . . . . . . . . . . . 43 91 5.1. Meshed CDNs . . . . . . . . . . . . . . . . . . . . . . . 43 92 5.2. CSP combined with CDN . . . . . . . . . . . . . . . . . . 44 93 5.3. CSP using CDNI Request Routing Interface . . . . . . . . 45 94 5.4. CDN Federations and CDN Exchanges . . . . . . . . . . . . 46 95 6. Trust Model . . . . . . . . . . . . . . . . . . . . . . . . . 49 96 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 50 97 8. Security Considerations . . . . . . . . . . . . . . . . . . . 50 98 8.1. Security of CDNI Interfaces . . . . . . . . . . . . . . . 51 99 8.2. Digital Rights Management . . . . . . . . . . . . . . . . 52 100 9. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 52 101 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 52 102 11. Informative References . . . . . . . . . . . . . . . . . . . 52 103 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 54 105 1. Introduction 107 This document provides an overview of the various components 108 necessary to interconnect CDNs, expanding on the problem statement 109 and use cases introduced in [RFC6770] and [RFC6707]. It describes 110 the necessary interfaces and mechanisms in general terms and outlines 111 how they fit together to form a complete system for CDN 112 Interconnection. Detailed specifications are left to other 113 documents. This document makes extensive use of message flow 114 examples to illustrate the operation of interconnected CDNs, but 115 these examples should be considered illustrative rather than 116 prescriptive. 118 [RFC3466] uses different terminology and models for "Content 119 Internetworking (CDI)". It is also less prescriptive in terms of 120 interfaces. To avoid confusion, this document obsoletes [RFC3466]. 122 1.1. Terminology 124 This document uses the core terminology defined in [RFC6707]. It 125 also introduces the following terms: 127 CDN-Domain: a host name (FQDN) at the beginning of a URL, 128 representing a set of content that is served by a given CDN. For 129 example, in the URL http://cdn.csp.example/...rest of url..., the CDN 130 domain is cdn.csp.example. A major role of CDN-Domain is to identify 131 a region (subset) of the URI space relative to which various CDN 132 Interconnection rules and policies are to apply. For example, a 133 record of CDN Metadata might be defined for the set of resources 134 corresponding to some CDN-Domain. 136 Distinguished CDN-Domain: a CDN-Domain that is allocated by a CDN for 137 the purposes of communication with a peer CDN, but which is not found 138 in client requests. Such CDN-Domains may be used for inter-CDN 139 acquisition, or as redirection targets, and enable a CDN to 140 distinguish a request from a peer CDN from an end-user request. 142 Delivering CDN: the CDN that ultimately delivers a piece of content 143 to the end-user. The last in a potential sequence of downstream 144 CDNs. 146 Recursive CDNI Request Redirection: When an upstream CDN elects to 147 redirect a request towards a downstream CDN, the upstream CDN can 148 query the downstream CDN Request Routing system via the CDNI Request 149 Routing Redirection Interface (or use information cached from earlier 150 similar queries) to find out how the downstream CDN wants the request 151 to be redirected, which allows the upstream CDN to factor in the 152 downstream CDN response when redirecting the user agent. This 153 approach is referred to as "Recursive" CDNI Request Redirection. 154 Note that the downstream CDN may elect to have the request redirected 155 directly to a Surrogate inside the downstream CDN, to the Request 156 Routing System of the downstream CDN, to another CDN, or to whatever 157 system is necessary to handle the redirected request appropriately. 159 Iterative CDNI Request Redirection: When an upstream CDN elects to 160 redirect a request towards a downstream CDN, the upstream CDN can 161 base its redirection purely on a local decision (and without 162 attempting to take into account how the downstream CDN may in turn 163 redirect the user agent). In that case, the upstream CDN redirects 164 the request to the request routing system in the downstream CDN, 165 which in turn will decide how to redirect that request: this approach 166 is referred to as "Iterative" CDNI Request Redirection. 168 Synchronous CDNI operations: operations between CDNs that happen 169 during the process of servicing a user request, i.e. between the time 170 that the user agent begins its attempt to obtain content and the time 171 at which that request is served. 173 Asynchronous CDNI operations: operations between CDNs that happen 174 independently of any given user request, such as advertisement of 175 footprint information or pre-positioning of content for later 176 delivery. 178 Trigger Interface: a subset of the CDNI Control interface that 179 includes operations to pre-position, revalidate, and purge both 180 metadata and content. These operations are typically called in 181 response to some action (Trigger) by the CSP on the upstream CDN. 183 We also sometimes use uCDN and dCDN as shorthand for upstream CDN and 184 downstream CDN, respectively. 186 1.2. Reference Model 188 This document uses the reference model in Figure 1, which expands the 189 reference model originally defined in [RFC6707]. (The difference is 190 that the expanded model splits the Request Routing Interface into its 191 two distinct parts: the Request Routing Redirection interface and the 192 Footprint and Capabilities Advertisement interface, as described 193 below.) 194 -------- 195 / \ 196 | CSP | 197 \ / 198 -------- 199 * 200 * 201 * /\ 202 * / \ 203 ---------------------- |CDNI| ---------------------- 204 / Upstream CDN \ | | / Downstream CDN \ 205 | +-------------+ | | CI | | +-------------+ | 206 |******* Control |<======|====|=======>| Control *******| 207 |* +------*----*-+ | | | | +-*----*------+ *| 208 |* * * | | | | * * *| 209 |* +------*------+ | | LI | | +------*------+ *| 210 |* ***** Logging |<======|====|=======>| Logging ***** *| 211 |* * +-*-----------+ | | | | +-----------*-+ * *| 212 |* * * * | | | | * * * *| 213 .....*...+-*---------*-+ | | RI | | +-*---------*-+...*.*... 214 . |* * | |<======|====|=======>| | * *| . 215 . |* * | Req-Routing | | |FCI | | | Req-Routing | * *| . 216 . |* * *** |<======|====|=======>| |** * *| . 217 . |* * * +-------------+.| | | | +-------------+ * * *| . 218 . |* * * . | | | * * *| . 219 . |* * * +-------------+ |. | MI | | +-------------+ * * *| . 220 . |* * * | Distribution|<==.===|====|=======>| Distribution| * * *| . 221 . |* * * | | | . \ / | | | * * *| . 222 . |* * * |+---------+ | | . \/ | | +---------+| * * *| . 223 . |* * ***| +---------+| | ...Request......+---------+ |*** * *| . 224 . |* *****+-|Surrogate|***********************|Surrogate|-+***** *| . 225 . |******* +---------+| | Acquisition | |+----------+ *******| . 226 . | +-------------+ | | +-------*-----+ | . 227 . \ / \ * / . 228 . ---------------------- ---------*------------ . 229 . * . 230 . * Delivery . 231 . * . 232 . +--*---+ . 233 ...............Request............................| User |..Request.. 234 | Agent| 235 +------+ 237 <==> interfaces inside the scope of CDNI 239 **** and .... interfaces outside the scope of CDNI 241 Figure 1: CDNI Expanded Model and CDNI Interfaces 243 We note that while some interfaces in the reference model are "out of 244 scope" for the CDNI WG (in the sense that there is no need to define 245 new protocols for those interfaces) we still need to refer to them in 246 this document to explain the overall operation of CDNI. 248 We also note that, while we generally show only one upstream CDN 249 serving a given CSP, it is entirely possible that multiple uCDNs can 250 serve a single CSP. In fact, this situation effectively exists today 251 in the sense that a single CSP can currently delegate its content 252 delivery to more than one CDN. 254 The following briefly describes the five CDNI interfaces, 255 paraphrasing the definitions given in [RFC6707]. We discuss these 256 interfaces in more detail in Section 4. 258 o CDNI Control interface (CI): Operations to bootstrap and 259 parameterize the other CDNI interfaces, as well as operations to 260 pre-position, revalidate, and purge both metadata and content. 261 The latter subset of operations is sometimes collectively called 262 the "Trigger interface." 264 o CDNI Request Routing interface: Operations to determine what CDN 265 (and optionally what surrogate within a CDN) is to serve end- 266 user's requests. This interface is actually a logical bundling of 267 two separate but related interfaces: 269 * CDNI Footprint & Capabilities Advertisement interface (FCI): 270 Asynchronous operations to exchange routing information (e.g., 271 the network footprint and capabilities served by a given CDN) 272 that enables CDN selection for subsequent user requests; and 274 * CDNI Request Routing Redirection interface (RI): Synchronous 275 operations to select a delivery CDN (surrogate) for a given 276 user request. 278 o CDNI Metadata interface (MI): Operations to communicate metadata 279 that governs how the content is delivered by interconnected CDNs. 280 Examples of CDNI metadata include geo-blocking directives, 281 availability windows, access control mechanisms, and purge 282 directives. It may include a combination of: 284 * Asynchronous operations to exchange metadata that govern 285 subsequent user requests for content; and 287 * Synchronous operations that govern behavior for a given user 288 request for content. 290 o CDNI Logging interface (LI): Operations that allow interconnected 291 CDNs to exchange relevant activity logs. It may include a 292 combination of: 294 * Real-time exchanges, suitable for runtime traffic monitoring; 295 and 297 * Offline exchanges, suitable for analytics and billing. 299 There is some potential overlap between the set of Trigger-based 300 operations in the CDNI Control interface and the CDNI Metadata 301 interface. For both cases, the information passed from the upstream 302 CDN to the downstream CDN can broadly be viewed as metadata that 303 describes how content is to be managed by the downstream CDN. For 304 example, the information conveyed by CI to pre-position, revalidate 305 or purge metadata is similar to the information conveyed by posting 306 updated metadata via the MI. Even the CI operation to purge content 307 could be viewed as an metadata update for that content: purge simply 308 says that the availability window for the named content ends now. 309 The two interfaces share much in common, so minimally, there will 310 need to be a consistent data model that spans both. 312 The distinction we draw has to do with what the uCDN knows about the 313 successful application of the metadata by the dCDN. In the case of 314 the CI, the downstream CDN returning a successful status message 315 guarantees that the operation has been successfully completed; e.g., 316 the content has been purged or pre-positioned. This implies that the 317 downstream CDN accepts responsibility for having successfully 318 completed the requested operation. In contrast, metadata passed 319 between CDNs via the MI carries no such completion guarantee. 320 Returning success implies successful receipt of the metadata, but 321 nothing can be inferred about precisely when the metadata will take 322 effect in the downstream CDN, only that it will take effect 323 eventually. This is because of the challenge in globally 324 synchronizing updates to metadata with end-user requests that are 325 currently in progress (or indistinguishable from currently being in 326 progress). Clearly, a CDN will not be viewed as a trusted peer if 327 "eventually" often becomes an indefinite period of time, but the 328 acceptance of responsibility cannot be as crisply defined for the MI. 330 Finally, there is a practical issue that impacts all of the CNDI 331 interfaces, and that is whether or not to optimize CDNI for HTTP 332 Adaptive Streaming (HAS). We highlight specific issues related to 333 delivering HAS content throughout this document, but for a more 334 thorough treatment of the topic, see [RFC6983]. 336 1.3. Structure Of This Document 338 The remainder of this document is organized as follows: 340 o Section 2 describes some essential building blocks for CDNI, 341 notably the various options for redirecting user requests to a 342 given CDN. 344 o Section 3 provides a number of illustrative examples of various 345 CDNI operations. 347 o Section 4 describes the functionality of the main CDNI interfaces. 349 o Section 5 shows how various deployment models of CDNI may be 350 achieved using the defined interfaces. 352 o Section 6 describes the trust model of CDNI and the issues of 353 transitive trust in particular that CDNI raises. 355 2. Building Blocks 357 2.1. Request Redirection 359 At its core, CDN Interconnection requires the redirection of requests 360 from one CDN to another. For any given request that is received by 361 an upstream CDN, it will either respond to the request directly, or 362 somehow redirect the request to a downstream CDN. Two main 363 mechanisms are available for redirecting a request to a downstream 364 CDN. The first leverages the DNS name resolution process and the 365 second uses in-protocol redirection mechanisms such as the HTTP 302 366 or 307 redirection response. We discuss these below as background 367 before discussing some examples of their use in Section 3. 369 2.1.1. DNS Redirection 371 DNS redirection is based on returning different IP addresses for the 372 same DNS name, for example, to balance server load or to account for 373 the client's location in the network. A DNS server, sometimes called 374 the Local DNS (LDNS), resolves DNS names on behalf of an end-user. 375 The LDNS server in turn queries other DNS servers until it reaches 376 the authoritative DNS server for the CDN-Domain. The network 377 operator typically provides the LDNS server, although the user is 378 free to choose other DNS servers (e.g., OpenDNS, Google Public DNS). 379 This latter possibility is important because the authoritative DNS 380 server sees only the IP address of the DNS server that queries it, 381 not the IP address of the original end-user. 383 The advantage of DNS redirection is that it is completely transparent 384 to the end user; the user sends a DNS name to the LDNS server and 385 gets back an IP address. On the other hand, DNS redirection is 386 problematic because the DNS request comes from the LDNS server, not 387 the end-user. This may affect the accuracy of server selection that 388 is based on the user's location. The transparency of DNS redirection 389 is also a problem in that there is no opportunity to take the 390 attributes of the user agent or the URI path component into account. 391 We consider two main forms of DNS redirection: simple and CNAME- 392 based. 394 In simple DNS redirection, the authoritative DNS server for the name 395 simply returns an IP address from a set of possible IP addresses. 396 The answer is chosen from the set based on characteristics of the set 397 (e.g., the relative loads on the servers) or characteristics of the 398 client (e.g., the location of the client relative to the servers). 399 Simple redirection is straightforward. The only caveats are (1) 400 there is a limit to the number of alternate IP addresses a single DNS 401 server can manage; and (2) DNS responses are cached by downstream 402 servers so the TTL on the response must be set to an appropriate 403 value so as to preserve the fresheness of the redirection. 405 In CNAME-based DNS redirection, the authoritative server returns a 406 CNAME response to the DNS request, telling the LDNS server to restart 407 the name lookup using a new name. A CNAME is essentially a symbolic 408 link in the DNS namespace, and like a symbolic link, redirection is 409 transparent to the client; the LDNS server gets the CNAME response 410 and re-executes the lookup. Only when the name has been resolved to 411 an IP address does it return the result to the user. Note that DNAME 412 would be preferable to CNAME if it becomes widely supported. 414 2.1.2. HTTP Redirection 416 HTTP redirection makes use of the redirection response of the HTTP 417 protocol (e.g.,"302" or "307"). This response contains a new URL 418 that the application should fetch instead of the original URL. By 419 changing the URL appropriately, the server can cause the user to 420 redirect to a different server. The advantages of HTTP redirection 421 are that (1) the server can change the URL fetched by the client to 422 include, for example, both the DNS name of the particular server to 423 use, as well as the original HTTP server that was being accessed; (2) 424 the client sends the HTTP request to the server, so that its IP 425 address is known and can be used in selecting the server; and (3) 426 other attributes (e.g., content type, user agent type) are visible to 427 the redirection mechanism. 429 The disadvantages of HTTP redirection are (1) it is visible to the 430 application, so it requires application support and may affect the 431 application behavior (e.g., web browsers will not send cookies if the 432 URL changes to a different domain); (2) HTTP is a heavyweight 433 protocol layered on TCP so it has relatively high overhead; and (3) 434 the results of HTTP redirection are not cached so that all 435 redirections must go through to the server. 437 3. Overview of CDNI Operation 439 To provide a big picture overview of the various components of CDN 440 Interconnection, we walk through a "day in the life" of a content 441 item that is made available via a pair of interconnected CDNs. This 442 will serve to illustrate many of the functions that need to be 443 supported in a complete CDNI solution. We give examples using both 444 DNS-based and HTTP-based redirection. We begin with very simple 445 examples and then how additional capabilities, such as recursive 446 request redirection and content removal, might be added. 448 Before walking through some specific examples, we present a high- 449 level view of the operations that may take place. This high-level 450 overview is illustrated in Figure 2. Note that most operations will 451 involve only a subset of all the messages shown below, and that the 452 order and number of operations may vary considerably, as more 453 detailed examples illustrate below. 455 The following shows Operator A as the upstream CDN (uCDN) and 456 Operator B as the downstream CDN (dCDN), where the former has a 457 relationship with a content provider and the latter being the CDN 458 selected by Operator A to deliver content to the end-user. The 459 interconnection relationship may be symmetric between these two CDN 460 operators, but each direction can be considered as operating 461 independently of the other so for simplicity we show the interaction 462 in one direction only. 464 End-User Operator B Operator A 465 | | | 466 | | | 467 | | [Async FCI Push] | (1) 468 | | | 469 | | [MI pre-positioning] | (2) 470 | | | 471 | CONTENT REQUEST | | 472 |-------------------------------------------------->| (3) 473 | | | 474 | | [Sync RI Pull] | (4) 475 | | | 476 | RI REPLY | | 477 |<--------------------------------------------------| (5) 478 | | | 479 | | | 480 | CONTENT REQUEST | | 481 |------------------------>| | (6) 482 | | | 483 | | [Sync MI Pull] | (7) 484 | | | 485 | | ACQUISITION REQUEST | 486 | X------------------------>| (8) 487 | X | 488 | X CONTENT DATA | 489 | X<------------------------| (9) 490 | | | 491 | CONTENT DATA | | 492 |<------------------------| | (10) 493 | | | 494 : : : 495 : [Other content requests] : 496 : : : 497 | | [CI: Content Purge] | (11) 498 : : : 499 | | [LI: Log exchange] | (12) 500 | | | 502 Figure 2: Overview of Operation 504 The operations shown in the Figure are as follows: 506 1. dCDN uses the FCI to advertise information relevant to its 507 delivery footprint and capabilities prior to any content 508 requests being redirected. 510 2. Prior to any content request, the uCDN uses the MI to 511 pre=position CDNI metadata to the dCDN, thereby making that 512 metadata available in readiness for later content requests. 514 3. A content request from a user agent arrives at uCDN. 516 4. uCDN may use the RI to synchronously request information from 517 dCDN regarding its delivery capabilities to decide if dCDN is a 518 suitable target for redirection of this request. 520 5. uCDN redirects the request to dCDN by sending some response 521 (DNS, HTTP) to the user agent. 523 6. The user agent requests the content from dCDN. 525 7. dCDN may use the MI to synchronously request metadata related to 526 this content from uCDN, e.g. to decide whether to serve it. 528 8. If the content is not already in a suitable cache in dCDN, dCDN 529 may acquire it from uCDN. 531 9. The content is delivered to dCDN from uCDN. 533 10. The content is delivered to the user agent by dCDN. 535 11. Some time later, perhaps at the request of the CSP (not shown) 536 uCDN may use the CI to instruct dCDN to purge the content, 537 thereby ensuring it is not delivered again. 539 12. After one or more content delivery actions by dCDN, a log of 540 delivery actions may be provided to uCDN using the LI. 542 The following sections show some more specific examples of how these 543 operations may be combined to perform various delivery, control and 544 logging operations across a pair of CDNs. 546 3.1. Preliminaries 548 Initially, we assume that there is at least one CSP that has 549 contracted with an upstream CDN (uCDN) to deliver content on its 550 behalf. We are not particularly concerned with the interface between 551 the CSP and uCDN, other than to note that it is expected to be the 552 same as in the "traditional" (non-interconnected) CDN case. Existing 553 mechanisms such as DNS CNAMEs or HTTP redirects (Section 2) can be 554 used to direct a user request for a piece of content from the CSP 555 towards the CSP's chosen upstream CDN. 557 We assume Operator A provides an upstream CDN that serves content on 558 behalf of a CSP with CDN-Domain cdn.csp.example. We assume that 559 Operator B provides a downstream CDN. An end user at some point 560 makes a request for URL 562 http://cdn.csp.example/...rest of url... 564 It may well be the case that cdn.csp.example is just a CNAME for some 565 other CDN-Domain (such as csp.op-a.example). Nevertheless, the HTTP 566 request in the examples that follow is assumed to be for the example 567 URL above. 569 Our goal is to enable content identified by the above URL to be 570 served by the CDN of operator B. In the following sections we will 571 walk through some scenarios in which content is served, as well as 572 other CDNI operations such as the removal of content from a 573 downstream CDN. 575 3.2. Iterative HTTP Redirect Example 577 In this section we walk through a simple, illustrative example using 578 HTTP redirection from uCDN to dCDN. The example also assumes the use 579 of HTTP redirection inside uCDN and dCDN; however, this is 580 independent of the choice of redirection approach across CDNs, so an 581 alternative example could be constructed still showing HTTP 582 redirection from uCDN to dCDN but using DNS for handling of request 583 inside each CDN. 585 We assume for this example that Operators A and B have established an 586 agreement to interconnect their CDNs, with A being upstream and B 587 being downstream. 589 The operators agree that a CDN-Domain peer-a.op-b.example will be 590 used as the target of redirections from uCDN to dCDN. We assume the 591 name of this domain is communicated by some means to each CDN. (This 592 could be established out-of-band or via a CDNI interface.) We refer 593 to this domain as a "distinguished" CDN-Domain to convey the fact 594 that its use is limited to the interconnection mechanism; such a 595 domain is never used directly by a CSP. 597 We assume the operators also agree on some distinguished CDN-Domain 598 that will be used for inter-CDN acquisition of CSP's content from 599 uCDN by dCDN. In this example, we'll use op-b-acq.op-a.example. 601 We assume the operators also exchange information regarding which 602 requests dCDN is prepared to serve. For example, dCDN may be 603 prepared to serve requests from clients in a given geographical 604 region or a set of IP address prefixes. This information may again 605 be provided out of band or via a defined CDNI interface. 607 We assume DNS is configured in the following way: 609 o The content provider is configured to make operator A the 610 authoritative DNS server for cdn.csp.example (or to return a CNAME 611 for cdn.csp.example for which operator A is the authoritative DNS 612 server). 614 o Operator A is configured so that a DNS request for 615 op-b-acq.op-a.example returns a request router in Operator A. 617 o Operator B is configured so that a DNS request for 618 peer-a.op-b.example/cdn.csp.example returns a request router in 619 Operator B. 621 Figure 3 illustrates how a client request for 623 http://cdn.csp.example/...rest of url... 625 is handled. 627 End-User Operator B Operator A 628 |DNS cdn.csp.example | | 629 |-------------------------------------------------->| 630 | | |(1) 631 |IPaddr of A's Request Router | 632 |<--------------------------------------------------| 633 |HTTP cdn.csp.example | | 634 |-------------------------------------------------->| 635 | | |(2) 636 |302 peer-a.op-b.example/cdn.csp.example | 637 |<--------------------------------------------------| 638 |DNS peer-a.op-b.example | | 639 |------------------------>| | 640 | |(3) | 641 |IPaddr of B's Request Router | 642 |<------------------------| | 643 | | | 644 |HTTP peer-a.op-b.example/cdn.csp.example | 645 |------------------------>| | 646 | |(4) | 647 |302 node1.peer-a.op-b.example/cdn.csp.example | 648 |<------------------------| | 649 |DNS node1.peer-a.op-b.example | 650 |------------------------>| | 651 | |(5) | 652 |IPaddr of B's Delivery Node | 653 |<------------------------| | 654 | | | 655 |HTTP node1.peer-a.op-b.example/cdn.csp.example | 656 |------------------------>| | 657 | |(6) | 658 | |DNS op-b-acq.op-a.example| 659 | |------------------------>| 660 | | |(7) 661 | |IPaddr of A's Request Router 662 | |<------------------------| 663 | |HTTP op-b-acq.op-a.example 664 | |------------------------>| 665 | | |(8) 666 | |302 node2.op-b-acq.op-a.example 667 | |<------------------------| 668 | |DNS node2.op-b-acq.op-a.example 669 | |------------------------>| 670 | | |(9) 671 | |IPaddr of A's Delivery Node 672 | |<------------------------| 673 | | | 674 | |HTTP node2.op-b-acq.op-a.example 675 | |------------------------>| 676 | | |(10) 677 | |Data | 678 | |<------------------------| 679 |Data | | 680 |<------------------------| | 682 Figure 3: Message Flow for Iterative HTTP Redirection 684 The steps illustrated in the figure are as follows: 686 1. A DNS resolver for Operator A processes the DNS request for its 687 customer based on CDN-Domain cdn.csp.example. It returns the IP 688 address of a request router in Operator A. 690 2. A Request Router for Operator A processes the HTTP request and 691 recognizes that the end-user is best served by another CDN, 692 specifically one provided by Operator B, and so it returns a 302 693 redirect message for a new URL constructed by "stacking" 694 Operator B's distinguished CDN-Domain (peer-a.op-b.example) on 695 the front of the original URL. (Note that more complex URL 696 manipulations are possible, such as replacing the initial CDN- 697 Domain by some opaque handle.) 699 3. The end-user does a DNS lookup using Operator B's distinguished 700 CDN-Domain (peer-a.op-b.example). B's DNS resolver returns the 701 IP address of a request router for Operator B. Note that if 702 request routing within dCDN was performed using DNS instead of 703 HTTP redirection, B's DNS resolver would also behave as the 704 request router and directly return the IP address of a delivery 705 node. 707 4. The request router for Operator B processes the HTTP request and 708 selects a suitable delivery node to serve the end-user request, 709 and returns a 302 redirect message for a new URL constructed by 710 replacing the hostname with a subdomain of the Operator B's 711 distinguished CDN-Domain that points to the selected delivery 712 node. 714 5. The end-user does a DNS lookup using Operator B's delivery node 715 subdomain (node1.peer-a.op-b.example). B's DNS resolver returns 716 the IP address of the delivery node. 718 6. The end-user requests the content from B's delivery node. In 719 the case of a cache hit, steps 6, 7, 8, 9 and 10 below do not 720 happen, and the content data is directly returned by the 721 delivery node to the end-user. In the case of a cache miss, the 722 content needs to be acquired by dCDN from uCDN (not the CSP). 723 The distinguished CDN-Domain peer-a.op-b.example indicates to 724 dCDN that this content is to be acquired from uCDN; stripping 725 the CDN-Domain reveals the original CDN-Domain cdn.csp.example 726 and dCDN may verify that this CDN-Domain belongs to a known peer 727 (so as to avoid being tricked into serving as an open proxy). 728 It then does a DNS request for an inter-CDN acquisition CDN- 729 Domain as agreed above (in this case, op-b-acq.op-a.example). 731 7. Operator A's DNS resolver processes the DNS request and returns 732 the IP address of a request router in operator A. 734 8. The request router for Operator A processes the HTTP request 735 from Operator B delivery node. Operator A request router 736 recognizes that the request is from a peer CDN rather than an 737 end-user because of the dedicated inter-CDN acquisition domain 738 (op-b-acq.op-a.example). (Note that without this specially 739 defined inter-CDN acquisition domain, operator A would be at 740 risk of redirecting the request back to operator B, resulting in 741 an infinite loop). The request router for Operator A selects a 742 suitable delivery node in uCDN to serve the inter-CDN 743 acquisition request and returns a 302 redirect message for a new 744 URL constructed by replacing the hostname with a subdomain of 745 the Operator A's distinguished inter-CDN acquisition domain that 746 points to the selected delivery node. 748 9. Operator A DNS resolver processes the DNS request and returns 749 the IP address of the delivery node in operator A. 751 10. Operator B requests (acquires) the content from Operator A. 752 Although not shown, Operator A processes the rest of the URL: it 753 extracts information identifying the origin server, validates 754 that this server has been registered, and determines the content 755 provider that owns the origin server. It may also perform its 756 own content acquisition steps if needed before returning the 757 content to dCDN. 759 The main advantage of this design is that it is simple: each CDN need 760 only know the distinguished CDN-Domain for each peer, with the 761 upstream CDN "pushing" the downstream CDN-Domain onto the URL as part 762 of its redirect (step 2) and the downstream CDN "popping" its CDN- 763 Domain off the URL to expose a CDN-Domain that the upstream CDN can 764 correctly process. Neither CDN needs to be aware of the internal 765 structure of the other's URLs. Moreover, the inter-CDN redirection 766 is entirely supported by a single HTTP redirect; neither CDN needs to 767 be aware of the other's internal redirection mechanism (i.e., whether 768 it is DNS or HTTP based). 770 One disadvantage is that the end-user's browser is redirected to a 771 new URL that is not in the same domain of the original URL. This has 772 implications on a number of security or validation mechanisms 773 sometimes used on endpoints. For example, it is important that any 774 redirected URL be in the same domain (e.g., csp.example) if the 775 browser is expected to send any cookies associated with that domain. 776 As another example, some video players enforce validation of a cross 777 domain policy that needs to allow for the domains involved in the CDN 778 redirection. These problems are generally soluble, but the solutions 779 complicate the example, so we do not discuss them further in this 780 version of the draft. 782 We note that this example begins to illustrate some of the interfaces 783 that may be required for CDNI, but does not require all of them. For 784 example, obtaining information from dCDN regarding the set of client 785 IP addresses or geographic regions it might be able to serve is an 786 aspect of request routing (specifically of the CDNI Footprint & 787 Capabilities Advertisement interface). Important configuration 788 information such as the distinguished names used for redirection and 789 inter-CDN acquisition could also be conveyed via a CDNI interface 790 (e.g., perhaps the CDNI Control interface). The example also shows 791 how existing HTTP-based methods suffice for the acquisition 792 interface. Arguably, the absolute minimum metadata required for CDNI 793 is the information required to acquire the content, and this 794 information was provided "in-band" in this example by means of the 795 URI handed to the client in the HTTP 302 response. The example also 796 assumes that the CSP does not require any distribution policy (e.g. 797 time window, geo-blocking) or delivery processing to be applied by 798 the interconnected CDNs. Hence, there is no explicit CDNI Metadata 799 interface invoked in this example. There is also no explicit CDNI 800 Logging interface discussed in this example. 802 We also note that the step of deciding when a request should be 803 redirected to dCDN rather than served by uCDN has been somewhat 804 glossed over. It may be as simple as checking the client IP address 805 against a list of prefixes, or it may be considerably more complex, 806 involving a wide range of factors, such as the geographic location of 807 the client (perhaps determined from a third party service), CDN load, 808 or specific business rules. 810 This example uses the "iterative" CDNI request redirection approach. 811 That is, uCDN performs part of the request redirection function by 812 redirecting the client to a request router in the dCDN, which then 813 performs the rest of the redirection function by redirecting to a 814 suitable surrogate. If request routing is performed in the dCDN 815 using HTTP redirection, this translates in the end-user experiencing 816 two successive HTTP redirections. By contrast, the alternative 817 approach of "recursive" CDNI request redirection effectively 818 coalesces these two successive HTTP redirections into a single one, 819 sending the end-user directly to the right delivery node in the dCDN. 820 This "recursive" CDNI request routing approach is discussed in the 821 next section. 823 3.3. Recursive HTTP Redirection Example 825 The following example builds on the previous one to illustrate the 826 use of the request routing interface (specifically the CDNI Request 827 Routing Redirection interface) to enable "recursive" CDNI request 828 routing. We build on the HTTP-based redirection approach because it 829 illustrates the principles and benefits clearly, but it is equally 830 possible to perform recursive redirection when DNS-based redirection 831 is employed. 833 In contrast to the prior example, the operators need not agree in 834 advance on a CDN-Domain to serve as the target of redirections from 835 uCDN to dCDN. We assume that the operators agree on some 836 distinguished CDN-Domain that will be used for inter-CDN acquisition 837 of CSP's content by dCDN. In this example, we'll use 838 op-b-acq.op-a.example. 840 We assume the operators also exchange information regarding which 841 requests dCDN is prepared to serve. For example, dCDN may be 842 prepared to serve requests from clients in a given geographical 843 region or a set of IP address prefixes. This information may again 844 be provided out of band or via a defined protocol. 846 We assume DNS is configured in the following way: 848 o The content provider is configured to make operator A the 849 authoritative DNS server for cdn.csp.example (or to return a CNAME 850 for cdn.csp.example for which operator A is the authoritative DNS 851 server). 853 o Operator A is configured so that a DNS request for 854 op-b-acq.op-a.example returns a request router in Operator A. 856 o Operator B is configured so that a request for node1.op-b.example/ 857 cdn.csp.example returns the IP address of a delivery node. Note 858 that there might be a number of such delivery nodes. 860 Figure 3 illustrates how a client request for 862 http://cdn.csp.example/...rest of url... 864 is handled. 866 End-User Operator B Operator A 867 |DNS cdn.csp.example | | 868 |-------------------------------------------------->| 869 | | |(1) 870 |IPaddr of A's Request Router | 871 |<--------------------------------------------------| 872 |HTTP cdn.csp.example | | 873 |-------------------------------------------------->| 874 | | |(2) 875 | |RR/RI REQ cdn.csp.example| 876 | |<------------------------| 877 | | | 878 | |RR/RI RESP node1.op-b.example 879 | |------------------------>| 880 | | |(3) 881 |302 node1.op-b.example/cdn.csp.example | 882 |<--------------------------------------------------| 883 |DNS node1.op-b.example | | 884 |------------------------>| | 885 | |(4) | 886 |IPaddr of B's Delivery Node | 887 |<------------------------| | 888 |HTTP node1.op-b.example/cdn.csp.example | 889 |------------------------>| | 890 | |(5) | 891 | |DNS op-b-acq.op-a.example| 892 | |------------------------>| 893 | | |(6) 894 | |IPaddr of A's Request Router 895 | |<------------------------| 896 | |HTTP op-b-acq.op-a.example 897 | |------------------------>| 898 | | |(7) 899 | |302 node2.op-b-acq.op-a.example 900 | |<------------------------| 901 | |DNS node2.op-b-acq.op-a.example 902 | |------------------------>| 903 | | |(8) 904 | |IPaddr of A's Delivery Node 905 | |<------------------------| 906 | | | 907 | |HTTP node2.op-b-acq.op-a.example 908 | |------------------------>| 909 | | |(9) 910 | |Data | 911 | |<------------------------| 912 |Data | | 913 |<------------------------| | 915 Figure 4: Message Flow for Recursive HTTP Redirection 917 The steps illustrated in the figure are as follows: 919 1. A DNS resolver for Operator A processes the DNS request for its 920 customer based on CDN-Domain cdn.csp.example. It returns the IP 921 address of a Request Router in Operator A. 923 2. A Request Router for Operator A processes the HTTP request and 924 recognizes that the end-user is best served by another CDN-- 925 specifically one provided by Operator B--and so it queries the 926 CDNI Request Routing Redirection interface of Operator B, 927 providing a set of information about the request including the 928 URL requested. Operator B replies with the DNS name of a 929 delivery node. 931 3. Operator A returns a 302 redirect message for a new URL obtained 932 from the RI. 934 4. The end-user does a DNS lookup using the host name of the URL 935 just provided (node1.op-b.example). B's DNS resolver returns the 936 IP address of the corresponding delivery node. Note that, since 937 the name of the delivery node was already obtained from B using 938 the RI, there should not be any further redirection here (in 939 contrast to the iterative method described above.) 941 5. The end-user requests the content from B's delivery node, 942 potentially resulting in a cache miss. In the case of a cache 943 miss, the content needs to be acquired from uCDN (not the CSP.) 944 The distinguished CDN-Domain op-b.example indicates to dCDN that 945 this content is to be acquired from another CDN; stripping the 946 CDN-Domain reveals the original CDN-Domain cdn.csp.example, dCDN 947 may verify that this CDN-Domain belongs to a known peer (so as to 948 avoid being tricked into serving as an open proxy). It then does 949 a DNS request for the inter-CDN Acquisition "distinguished" CDN- 950 Domain as agreed above (in this case, op-b-acq.op-a.example). 952 6. Operator A DNS resolver processes the DNS request and returns the 953 IP address of a request router in operator A. 955 7. The request router for Operator A processes the HTTP request from 956 Operator B delivery node. Operator A request router recognizes 957 that the request is from a peer CDN rather than an end-user 958 because of the dedicated inter-CDN acquisition domain 959 (op-b-acq.op-a.example). (Note that without this specially 960 defined inter-CDN acquisition domain, operator A would be at risk 961 of redirecting the request back to operator B, resulting in an 962 infinite loop). The request router for Operator A selects a 963 suitable delivery node in uCDN to serve the inter-CDN acquisition 964 request and returns a 302 redirect message for a new URL 965 constructed by replacing the hostname with a subdomain of the 966 Operator A's distinguished inter-CDN acquisition domain that 967 points to the selected delivery node. 969 8. Operator A recognizes that the DNS request is from a peer CDN 970 rather than an end-user (due to the internal CDN-Domain) and so 971 returns the address of a delivery node. (Note that without this 972 specially defined internal domain, Operator A would be at risk of 973 redirecting the request back to Operator B, resulting in an 974 infinite loop.) 976 9. Operator B requests (acquires) the content from Operator A. 977 Operator A serves content for the requested CDN-Domain to dCDN. 978 Although not shown, it is at this point that Operator A processes 979 the rest of the URL: it extracts information identifying the 980 origin server, validates that this server has been registered, 981 and determines the content provider that owns the origin server. 982 It may also perform its own content acquisition steps if needed 983 before returning the content to dCDN. 985 Recursive redirection has the advantage over iterative of being more 986 transparent from the end-user's perspective, but the disadvantage of 987 each CDN exposing more of its internal structure (in particular, the 988 addresses of edge caches) to peer CDNs. By contrast, iterative 989 redirection does not require dCDN to expose the addresses of its edge 990 caches to uCDN. 992 This example happens to use HTTP-based redirection in both CDN A and 993 CDN B, but a similar example could be constructed using DNS-based 994 redirection in either CDN. Hence, the key point to take away here is 995 simply that the end user only sees a single redirection of some type, 996 as opposed to the pair of redirections in the prior (iterative) 997 example. 999 The use of the RI requires that the request routing mechanism be 1000 appropriately configured and bootstrapped, which is not shown here. 1001 More discussion on the bootstrapping of interfaces is provided in 1002 Section 4 1004 3.4. Iterative DNS-based Redirection Example 1006 In this section we walk through a simple example using DNS-based 1007 redirection for request redirection from uCDN to dCDN (as well as for 1008 request routing inside dCDN and uCDN). As noted in Section 2.1, DNS- 1009 based redirection has certain advantages over HTTP-based redirection 1010 (notably, it is transparent to the end-user) as well as some 1011 drawbacks (notably the client IP address is not visible to the 1012 request router). 1014 As before, Operator A has to learn the set of requests that dCDN is 1015 willing or able to serve (e.g. which client IP address prefixes or 1016 geographic regions are part of the dCDN footprint). We assume 1017 Operator has and makes known to operator A some unique identifier 1018 that can be used for the construction of a distinguished CDN-Domain, 1019 as shown in more detail below. (This identifier strictly needs only 1020 to be unique within the scope of Operator A, but a globally unique 1021 identifier, such as an AS number assigned to B, is one easy way to 1022 achieve that.) Also, Operator A obtains the NS records for Operator 1023 B's externally visible redirection servers. Also, as before, a 1024 distinguished CDN-Domain, such as op-b-acq.op-a.example, must be 1025 assigned for inter-CDN acquisition. 1027 We assume DNS is configured in the following way: 1029 o The CSP is configured to make Operator A the authoritative DNS 1030 server for cdn.csp.example (or to return a CNAME for 1031 cdn.csp.example for which operator A is the authoritative DNS 1032 server). 1034 o When uCDN sees a request best served by dCDN, it returns CNAME and 1035 NS records for "b.cdn.csp.example", where "b" is the unique 1036 identifier assigned to Operator B. (It may, for example, be an AS 1037 number assigned to Operator B.) 1039 o dCDN is configured so that a request for "b.cdn.csp.example" 1040 returns a delivery node in dCDN. 1042 o uCDN is configured so that a request for "op-b-acq.op-a.example" 1043 returns a delivery node in uCDN. 1045 Figure 5 depicts the exchange of DNS and HTTP requests. The main 1046 differences from Figure 3 are the lack of HTTP redirection and 1047 transparency to the end-user. 1049 End-User Operator B Operator A 1050 |DNS cdn.csp.example | | 1051 |-------------------------------------------------->| 1052 | | |(1) 1053 |CNAME b.cdn.csp.example | | 1054 |NS records for b.cdn.csp.example | 1055 |<--------------------------------------------------| 1056 |DNS b.cdn.csp.example | | 1057 |------------------------>| | 1058 | |(2) | 1059 |IPaddr of B's Delivery Node | 1060 |<------------------------| | 1061 |HTTP cdn.csp.example | | 1062 |------------------------>| | 1063 | |(3) | 1064 | |DNS op-b-acq.op-a.example| 1065 | |------------------------>| 1066 | | |(4) 1067 | |IPaddr of A's Delivery Node 1068 | |<------------------------| 1069 | |HTTP op-b-acq.op-a.example 1070 | |------------------------>| 1071 | | |(5) 1072 | |Data | 1073 | |<------------------------| 1074 |Data | | 1075 |<------------------------| | 1077 Figure 5: Message Flow for DNS-based Redirection 1079 The steps illustrated in the figure are as follows: 1081 1. Request Router for Operator A processes the DNS request for CDN- 1082 Domain cdn.csp.example and recognizes that the end-user is best 1083 served by another CDN. (This may depend on the IP address of the 1084 user's local DNS resolver, or other information discussed below.) 1085 The Request Router returns a DNS CNAME response by "stacking" the 1086 distinguished identifier for Operator B onto the original CDN- 1087 Domain (e.g., b.cdn.csp.example), plus an NS record that maps 1088 b.cdn.csp.example to B's Request Router. 1090 2. The end-user does a DNS lookup using the modified CDN-Domain 1091 (i.e., b.cdn.csp.example). This causes B's Request Router to 1092 respond with a suitable delivery node. 1094 3. The end-user requests the content from B's delivery node. The 1095 requested URL contains the name cdn.csp.example. (Note that the 1096 returned CNAME does not affect the URL.) At this point the 1097 delivery node has the correct IP address of the end-user and can 1098 do an HTTP 302 redirect if the redirections in steps 2 and 3 were 1099 incorrect. Otherwise B verifies that this CDN-Domain belongs to 1100 a known peer (so as to avoid being tricked into serving as an 1101 open proxy). It then does a DNS request for an "internal" CDN- 1102 Domain as agreed above (op-b-acq.op-a.example). 1104 4. Operator A recognizes that the DNS request is from a peer CDN 1105 rather than an end-user (due to the internal CDN-Domain) and so 1106 returns the address of a delivery node in uCDN. 1108 5. Operator A serves content to dCDN. Although not shown, it is at 1109 this point that Operator A processes the rest of the URL: it 1110 extracts information identifying the origin server, validates 1111 that this server has been registered, and determines the content 1112 provider that owns the origin server. 1114 The advantages of this approach are that it is more transparent to 1115 the end-user and requires fewer round trips than HTTP-based 1116 redirection (in its worst case, i.e., when none of the needed DNS 1117 information is cached). A potential problem is that the upstream CDN 1118 depends on being able to learn the correct downstream CDN that serves 1119 the end-user from the client address in the DNS request. In standard 1120 DNS operation, uCDN will only obtain the address of the client's 1121 local DNS resolver (LDNS), which is not guaranteed to be in the same 1122 network (or geographic region) as the client. If not--e.g., the end- 1123 user uses a global DNS service--then the upstream CDN cannot 1124 determine the appropriate downstream CDN to serve the end-user. In 1125 this case, and assuming the uCDN is capable of detecting that 1126 situation, one option is for the upstream CDN to treat the end-user 1127 as it would any user not connected to a peer CDN. Another option is 1128 for the upstream CDN to "fall back" to a pure HTTP-based redirection 1129 strategy in this case (i.e., use the first method). Note that this 1130 problem affects existing CDNs that rely on DNS to determine where to 1131 redirect client requests, but the consequences are arguably less 1132 serious for CDNI since the LDNS is likely in the same network as the 1133 dCDN serves. One approach to ensuring that the client's IP address 1134 prefix is correctly determined in such situations is described in 1135 [I-D.vandergaast-edns-client-subnet]. 1137 As with the prior example, this example partially illustrates the 1138 various interfaces involved in CDNI. Operator A could learn 1139 dynamically from Operator B the set of prefixes or regions that B is 1140 willing and able to serve via the CDNI Footprint & Capabilities 1141 Advertisement interface. The distinguished name used for acquisition 1142 and the identifier for Operator B that is prepended to the CDN-Domain 1143 on redirection are examples of information elements that might also 1144 be conveyed by CDNI interfaces (or, alternatively, statically 1145 configured). As before, minimal metadata sufficient to obtain the 1146 content is carried "in-band" as part of the redirection process, and 1147 standard HTTP is used for inter-CDN acquisition. There is no 1148 explicit CDNI Logging interface discussed in this example. 1150 3.5. Dynamic Footprint Discovery Example 1152 There could be situations where being able to dynamically discover 1153 the set of requests that a given dCDN is willing and able to serve is 1154 beneficial. For example, a CDN might at one time be able to serve a 1155 certain set of client IP prefixes, but that set might change over 1156 time due to changes in the topology and routing policies of the IP 1157 network. The following example illustrates this capability. We have 1158 chosen the example of DNS-based redirection, but HTTP-based 1159 redirection could equally well use this approach. 1161 End-User Operator B Operator A 1162 |DNS cdn.csp.example | | 1163 |-------------------------------------------------->| 1164 | | |(1) 1165 | | RI REQ op-b.example | 1166 | |<------------------------| 1167 | | |(2) 1168 | | RI REPLY | 1169 | |------------------------>| 1170 | | |(3) 1171 |CNAME b.cdn.csp.example | | 1172 |NS records for b.cdn.csp.example | 1173 |<--------------------------------------------------| 1174 |DNS b.cdn.csp.example | | 1175 |------------------------>| | 1176 | |(2) | 1177 |IPaddr of B's Delivery Node | 1178 |<------------------------| | 1179 |HTTP cdn.csp.example | | 1180 |------------------------>| | 1181 | |(3) | 1182 | |DNS op-b-acq.op-a.example| 1183 | |------------------------>| 1184 | | |(4) 1185 | |IPaddr of A's Delivery Node 1186 | |<------------------------| 1187 | |HTTP op-b-acq.op-a.example 1188 | |------------------------>| 1189 | | |(5) 1190 | |Data | 1191 | |<------------------------| 1192 |Data | | 1193 |<------------------------| | 1195 Figure 6: Message Flow for Dynamic Footprint Discovery 1197 This example differs from the one in Figure 5 only in the addition of 1198 a FCI request (step 2) and corresponding response (step 3). The RI 1199 REQ could be a message such as "Can you serve clients from this IP 1200 Prefix?" or it could be "Provide the list of client IP prefixes you 1201 can currently serve". In either case the response might be cached by 1202 operator A to avoid repeatedly asking the same question. 1203 Alternatively, or in addition, Operator B may spontaneously advertise 1204 to Operator A information (or changes) on the set of requests it is 1205 willing and able to serve on behalf of operator A; in that case, 1206 Operator B may spontaneously issue RR/RI REPLY messages that are not 1207 in direct response to a corresponding RR/RI REQ message. (Note that 1208 the issues of determining the client's subnet from DNS requests, as 1209 described above, are exactly the same here as in Section 3.4.) 1211 Once Operator A obtains the RI response, it is now able to determine 1212 that Operator B's CDN is an appropriate dCDN for this request and 1213 therefore a valid candidate dCDN to consider in its Redirection 1214 decision. If that dCDN is selected, the redirection and serving of 1215 the request proceeds as before (i.e. in the absence of dynamic 1216 footprint discovery). 1218 3.6. Content Removal Example 1220 The following example illustrates how the CDNI Control interface may 1221 be used to achieve pre-positioning of an item of content in the dCDN. 1222 In this example, user requests for a particular content, and 1223 corresponding redirection of such requests from Operator A to 1224 Operator B CDN, may (or may not) have taken place earlier. Then, at 1225 some point in time, the uCDN (for example, in response to a 1226 corresponding Trigger from the Content Provider) uses the CI to 1227 request that content identified by a particular URL be removed from 1228 dCDN. The following diagram illustrates the operation. 1230 End-User Operator B Operator A 1231 | |CI purge cdn.csp.example/... 1232 | |<------------------------| 1233 | | | 1234 | |CI OK | 1235 | |------------------------>| 1236 | | | 1238 Figure 7: Message Flow for Content Removal 1240 The CI is used to convey the request from uCDN to dCDN that some 1241 previously acquired content should be deleted. The URL in the 1242 request specifies which content to remove. This example corresponds 1243 to a DNS-based redirection scenario such as Section 3.4. If HTTP- 1244 based redirection had been used, the URL for removal would be of the 1245 form peer-a.op-b.example/cdn.csp.example/... 1247 The dCDN is expected to confirm to the uCDN, as illustrated by the CI 1248 OK message, the completion of the removal of the targeted content 1249 from all the caches in dCDN. 1251 3.7. Pre-Positioned Content Acquisition Example 1253 The following example illustrates how the CI may be used to pre- 1254 position an item of content in the dCDN. In this example, Operator A 1255 uses the CDNI Metadata interface to request that content identified 1256 by a particular URL be pre-positioned into Operator B CDN. 1258 End-User Operator B Operator A 1260 | |CI pre-position cdn.csp.example/... 1261 | |<------------------------| 1262 | | |(1) 1263 | |CI OK | 1264 | |------------------------>| 1265 | | | 1266 | |DNS op-b-acq.op-a.example| 1267 | |------------------------>| 1268 | | |(2) 1269 | |IPaddr of A's Delivery Node 1270 | |<------------------------| 1271 | |HTTP op-b-acq.op-a.example 1272 | |------------------------>| 1273 | | |(3) 1274 | |Data | 1275 | |<------------------------| 1276 |DNS cdn.csp.example | | 1277 |--------------------------------------------->| 1278 | | |(4) 1279 |IPaddr of A's Request Router | 1280 |<---------------------------------------------| 1281 |HTTP cdn.csp.example| | 1282 |--------------------------------------------->| 1283 | | |(5) 1284 |302 peer-a.op-b.example/cdn.csp.example | 1285 |<---------------------------------------------| 1286 |DNS peer-a.op-b.example | 1287 |------------------->| | 1288 | |(6) | 1289 |IPaddr of B's Delivery Node | 1290 |<-------------------| | 1291 |HTTP peer-a.op-b.example/cdn.csp.example | 1292 |------------------->| | 1293 | |(7) | 1294 |Data | | 1295 |<-------------------| | 1297 Figure 8: Message Flow for Content Pre-Positioning 1299 The steps illustrated in the figure are as follows: 1301 1. Operator A uses the CI to request that Operator B pre-positions a 1302 particular content item identified by its URL. Operator B 1303 responds by confirming that it is willing to perform this 1304 operation. 1306 Steps 2 and 3 are exactly the same as steps 5 and 6 of Figure 3, only 1307 this time those steps happen as the result of the Pre-positioning 1308 request instead of as the result of a cache miss. 1310 Steps 4, 5, 6, 7 are exactly the same as steps 1, 2, 3, 4 of 1311 Figure 3, only this time Operator B CDN can serve the end-user 1312 request without triggering dynamic content acquisition, since the 1313 content has been pre-positioned in dCDN. Note that, depending on 1314 dCDN operations and policies, the content pre-positioned in the dCDN 1315 may be pre-positioned to all, or a subset of, dCDN caches. In the 1316 latter case, intra-CDN dynamic content acquisition may take place 1317 inside the dCDN serving requests from caches on which the content has 1318 not been pre-positioning; however, such intra-CDN dynamic acquisition 1319 would not involve the uCDN. 1321 3.8. Asynchronous CDNI Metadata Example 1323 In this section we walk through a simple example illustrating a 1324 scenario of asynchronously exchanging CDNI metadata, where the 1325 downstream CDN obtains CDNI metadata for content ahead of a 1326 corresponding content request. The example that follows assumes that 1327 HTTP-based inter-CDN redirection and recursive CDNI request-routing 1328 are used, as in Section 3.3. However, Asynchronous exchange of CDNI 1329 Metadata is similarly applicable to DNS-based inter-CDN redirection 1330 and iterative request routing (in which cases the CDNI metadata may 1331 be used at slightly different processing stages of the message 1332 flows). 1334 End-User Operator B Operator A 1335 | | | 1336 | |CI pre-position (Trigger)| 1337 | |<------------------------|(1) 1338 | | | 1339 | |CI OK | 1340 | |------------------------>|(2) 1341 | | | 1342 | |MI pull REQ | 1343 | |------------------------>|(3) 1344 | | | 1345 | |MI metadata REP |(4) 1346 | | | 1347 | | | 1348 | CONTENT REQUEST | | 1349 |-------------------------------------------------->|(5) 1350 | | | 1351 | | RI REQ | 1352 | |<------------------------|(6) 1353 | | | 1354 | | RI RESP | 1355 | |------------------------>|(7) 1356 | | | 1357 | CONTENT REDIRECTION | | 1358 |<--------------------------------------------------|(8) 1359 | | | 1360 | CONTENT REQUEST | | 1361 |------------------------>|(9) | 1362 | | | 1363 : : : 1364 | CONTENT DATA | | 1365 |<------------------------| |(10) 1367 Figure 9: Message Flow for Asynchronous CDNI Metadata 1369 The steps illustrated in the figure are as follows: 1371 1. Operator A uses the CI to Trigger to signal the availability of 1372 CDNI metadata to Operator B. 1374 2. Operator B acknowledges the receipt of this Trigger. 1376 3. Operator B requests the latest metadata from Operator A using 1377 the MI. 1379 4. Operator A replies with the requested metadata. This document 1380 does not constrain how the CDNI metadata information is actually 1381 represented. For the purposes of this example, we assume that 1382 Operator A provides CDNI metadata to Operator B indicating that: 1384 * this CDNI Metadata is applicable to any content referenced by 1385 some CDN-Domain. 1387 * this CDNI metadata consists of a distribution policy 1388 requiring enforcement by the delivery node of a specific per- 1389 request authorization mechanism (e.g. URI signature or token 1390 validation). 1392 5. A Content Request occurs as usual. 1394 6. A CDNI Request Routing Redirection request (RI REQ) is issued by 1395 operator A CDN, as discussed in Section 3.3. Operator B's 1396 request router can access the CDNI Metadata that are relevant to 1397 the requested content and that have been pre-positioned as per 1398 Steps 1-4, which may or may not affect the response. 1400 7. Operator B's request router issues a CDNI Request Routing 1401 Redirection response (RI RESP) as in Section 3.3. 1403 8. Operator B performs content redirection as discussed in 1404 Section 3.3. 1406 9. On receipt of the Content Request by the end user, the delivery 1407 node detects that previously acquired CDNI metadata is 1408 applicable to the requested content. In accordance with the 1409 specific CDNI metadata of this example, the delivery node will 1410 invoke the appropriate per-request authorization mechanism, 1411 before serving the content. (Details of this authorization are 1412 not shown.) 1414 10. Assuming successful per-request authorization, serving of 1415 Content Data (possibly preceded by inter-CDN acquisition) 1416 proceeds as in Section 3.3. 1418 3.9. Synchronous CDNI Metadata Acquisition Example 1420 In this section we walk through a simple example illustrating a 1421 scenario of Synchronous CDNI metadata acquisition, in which the 1422 downstream CDN obtains CDNI metadata for content at the time of 1423 handling a first request for the corresponding content. As in the 1424 preceding section, this example assumes that HTTP-based inter-CDN 1425 redirection and recursive CDNI request-routing are used (as in 1426 Section 3.3), but dynamic CDNI metadata acquisition is applicable to 1427 other variations of request routing. 1429 End-User Operator B Operator A 1430 | | | 1431 | CONTENT REQUEST | | 1432 |-------------------------------------------------->|(1) 1433 | | | 1434 | | RI REQ | 1435 | (2)|<------------------------| 1436 | | | 1437 | | MI REQ | 1438 | (3)|------------------------>| 1439 | | MI RESP | 1440 | |<------------------------|(4) 1441 | | | 1442 | | RI RESP | 1443 | |------------------------>|(5) 1444 | | | 1445 | | | 1446 | CONTENT REDIRECTION | | 1447 |<--------------------------------------------------|(6) 1448 | | | 1449 | CONTENT REQUEST | | 1450 |------------------------>|(7) | 1451 | | | 1452 | | MI REQ | 1453 | (8)|------------------------>| 1454 | | MI RESP | 1455 | |<------------------------|(9) 1456 | | | 1457 : : : 1458 | CONTENT DATA | | 1459 |<------------------------| |(10) 1461 Figure 10: Message Flow for Synchronous CDNI Metadata Acquisition 1463 The steps illustrated in the figure are as follows: 1465 1. A Content Request arrives as normal. 1467 2. An RI request occurs as in the prior example. 1469 3. On receipt of the CDNI Request Routing Request, Operator B's CDN 1470 initiates Synchronous acquisition of CDNI Metadata that are 1471 needed for routing of the end-user request. We assume the URI 1472 for the a Metadata server is known ahead of time through some 1473 out-of-band means. 1475 4. On receipt of a CDNI Metadata Request, Operator A's CDN 1476 responds, making the corresponding CDNI metadata information 1477 available to Operator B's CDN. This metadata is considered by 1478 operator B's CDN before responding to the Request Routing 1479 request. (In a simple case, the metadata could simply be an 1480 allow or deny response for this particular request.) 1482 5. Response to the RI request as normal. 1484 6. Redirection message is sent to the end user. 1486 7. A delivery node of Operator B receives the end user request. 1488 8. The delivery node Triggers dynamic acquisition of additional 1489 CDNI metadata that are needed to process the end-user content 1490 request. Note that there may exist cases where this step need 1491 not happen, for example because the metadata were already 1492 acquired previously. 1494 9. Operator A's CDN responds to the CDNI Metadata Request and makes 1495 the corresponding CDNI metadata available to Operator B. This 1496 metadata influence how Operator B's CDN processes the end-user 1497 request. 1499 10. Content is served (possibly preceded by inter-CDN acquisition) 1500 as in Section 3.3. 1502 3.10. Content and Metadata Acquisition with Multiple Upstream CDNs 1504 A single dCDN may receive end-user requests from multiple uCDNs. 1505 When a dCDN receives an end-user request, it must determine the 1506 identity of the uCDN from which it should acquire the requested 1507 content. 1509 Ideally, the acquisition path of an end-user request will follow the 1510 redirection path of the request. The dCDN should acquire the content 1511 from the same uCDN which redirected the request. 1513 Determining the acquisition path requires the dCDN to reconstruct the 1514 redirection path based on information in the end-user request. The 1515 method for reconstructing the redirection path differs based on the 1516 redirection approach: HTTP or DNS. 1518 With HTTP-redirection, the rewritten URI should include sufficient 1519 information for the dCDN to directly or indirectly determine the uCDN 1520 when the end-user request is received. The HTTP-redirection approach 1521 can be further broken-down based on the how the URL is rewritten 1522 during redirection: HTTP-redirection with or without Site 1523 Aggregation. HTTP-redirection with Site Aggregation hides the 1524 identity of the original CSP. HTTP-redirection without Site 1525 Aggregation does not attempt to hide the identity of the original 1526 CSP. With both approaches, the rewritten URI includes enough 1527 information to identify the immediate neighbor uCDN. 1529 With DNS-redirection, the dCDN receives the published URI (instead of 1530 a rewritten URI) and does not have sufficient information for the 1531 dCDN to identify the appropriate uCDN. The dCDN may narrow the set 1532 of viable uCDNs by examining the CDNI metadata from each to determine 1533 which uCDNs are hosting metadata for the requested content. If there 1534 is a single uCDN hosting metadata for the requested content, the dCDN 1535 can assume that the request redirection is coming from this uCDN and 1536 can acquire content from that uCDN. If there are multiple uCDNs 1537 hosting metadata for the requested content, the dCDN may be ready to 1538 trust any of these uCDNs to acquire the content (provided the uCDN is 1539 in a position to serve it). If the dCDN is not ready to trust any of 1540 these uCDNs, it needs to ensure via out of band arrangements that, 1541 for a given content, only a single uCDN will ever redirect requests 1542 to the dCDN. 1544 Content acquisition may be preceded by content metadata acquisition. 1545 If possible, the acquisition path for metadata should also follow the 1546 redirection path. Additionally, we assume metadata is indexed based 1547 on rewritten URIs in the case of HTTP-redirection and is indexed 1548 based on published URIs in the case of DNS-redirection. Thus, the RI 1549 and the MI are tightly coupled in that the result of request routing 1550 (a rewritten URI pointing to the dCDN) serves as an input to metadata 1551 lookup. If the content metadata includes information for acquiring 1552 the content, then the MI is also tightly coupled with the acquisition 1553 interface in that the result of the metadata lookup (an acquisition 1554 URL likely hosted by the uCDN) should serve as input to the content 1555 acquisition. 1557 4. Main Interfaces 1559 Figure 1 illustrates the main interfaces that are in scope for the 1560 CDNI WG, along with several others. The detailed specifications of 1561 these interfaces are left to other documents, but see [RFC6707] and 1562 [I-D.ietf-cdni-requirements] for some discussion of the interfaces. 1564 One interface that is not shown in Figure 1 is the interface between 1565 the user and the CSP. While for the purposes of CDNI that interface 1566 is out of scope, it is worth noting that it does exist and can 1567 provide useful functions, such as end-to-end performance monitoring 1568 and some forms of authentication and authorization. 1570 There is also an important interface between the user and the Request 1571 Routing function of both uCDN and dCDN (shown as the "Request" 1572 Interface in Figure 1). As we saw in some of the preceding examples, 1573 that interface can be used as a way of passing information a subset 1574 of metadata such as the minimum information that is required for dCDN 1575 to obtain the content from uCDN. 1577 In this section we will provide an overview of the functions 1578 performed by each of the CDNI interfaces and discuss how they fit 1579 into the overall solution. We also examine some of the design 1580 tradeoffs, and explore several cross-interface concerns. We begin 1581 with an examination of one such tradeoff that affects all the 1582 interfaces - the use of in-band or out-of-band communication. 1584 4.1. In-Band versus Out-of-Band Interfaces 1586 Before getting to the individual interfaces, we observe that there is 1587 a high-level design choice for each, involving the use of existing 1588 in-band communication channels versus defining new out-of-band 1589 interfaces. 1591 It is possible that the information needed to carry out various 1592 interconnection functions can be communicated between peer CDNs using 1593 existing in-band protocols. The use of HTTP 302 redirect is an 1594 example of how certain aspects of request routing can be implemented 1595 in-band (embedded in URIs). Note that using existing in-band 1596 protocols does not imply that the CDNI interfaces are null; it is 1597 still necessary to establish the rules (conventions) by which such 1598 protocols are used to implement the various interface functions. 1600 There are other opportunities for in-band communication beyond HTTP 1601 redirects. For example, many of the HTTP directives used by proxy 1602 servers can also be used by peer CDNs to inform each other of caching 1603 activity. Of these, one that is particularly relevant is the If- 1604 Modified-Since directive, which is used with the GET method to make 1605 it conditional: if the requested object has not been modified since 1606 the time specified in this field, a copy of the object will not be 1607 returned, and instead, a 304 (not modified) response will be 1608 returned. 1610 4.2. Cross Interface Concerns 1612 Although the CDNI interfaces are largely independent, there are a set 1613 of conventions practiced consistently across all interfaces. Most 1614 important among these is how resources are named, for exampmle, how 1615 the CDNI Metadata and Control interfaces identify the set of 1616 resources to which a given directive applies, or the CDNI Logging 1617 interface identifies the set of resources for which a summary record 1618 applies. 1620 While in the limit the CDNI interfaces could explicitly identify 1621 every individual resource, in practice, they name resource aggregates 1622 (sets of URIs) that are to be treated in a similar way. For example, 1623 URI aggregates can be identified by a CDN-Domain (i.e., the FQDN at 1624 the beginning of a URI) or by a URI-Filter (i.e., a regular 1625 expression that matches a subset of URIs contained in some CDN- 1626 Doman). In other words, CDN-Domains and URI-Filters provide a 1627 uniform means to aggregate sets (and subsets) of URIs for the purpose 1628 of defining the scope for some operation in one of the CDNI 1629 interfaces. 1631 4.3. Request Routing Interfaces 1633 The Request Routing interface comprises two parts: the Asynchronous 1634 interface used by a dCDN to advertize footprint and capabilities 1635 (denoted FCI) to a uCDN, allowing the uCDN to decide whether to 1636 redirect particular user requests to that dCDN; and the Synchronous 1637 interface used by the uCDN to redirect a user request to the dCDN 1638 (denoted RI). (These are somewhat analogous to the operations of 1639 routing and forwarding in IP.) 1641 As illustrated in Section 3, the RI part of request routing may be 1642 implemented in part by DNS and HTTP. Naming conventions may be 1643 established by which CDN peers communicate whether a request should 1644 be routed or content served. 1646 We also note that RI plays a key role in enabling recursive 1647 redirection, as illustrated in Section 3.3. It enables the user to 1648 be redirected to the correct delivery node in dCDN with only a single 1649 redirection step (as seen by the user). This may be particularly 1650 valuable as the chain of interconnected CDNs increases beyond two 1651 CDNs. For further discussion on the RI, see 1652 [I-D.ietf-cdni-redirection]. 1654 In support of these redirection requests, it is necessary for CDN 1655 peers to exchange additional information with each other, and this is 1656 the role of the FCI part of request routing. Depending on the 1657 method(s) supported, this might includes 1659 o The operator's unique id (operator-id) or distinguished CDN-Domain 1660 (operator-domain); 1662 o NS records for the operator's set of externally visible request 1663 routers; 1665 o The set of requests the dCDN operator is prepared to serve (e.g. a 1666 set of client IP prefixes or geographic regions that may be served 1667 by dCDN). 1669 o Additional capabilities of the dCDN, such as its ability to 1670 support different CDNI Metadata requests. 1672 Note that the set of requests that dCDN is willing to serve could in 1673 some cases be relatively static (e.g., a set of IP prefixes) which 1674 could be exchanged off-line, or might even be negotiated as part of a 1675 peering agreement. However, it may also be more dynamic, in which 1676 case the exchange supported by FCI would be be helpful. A further 1677 discussion of the Footprint & Capability Advertisement interface can 1678 be found in [I-D.ietf-cdni-footprint-capabilities-semantics]. 1680 4.4. CDNI Logging Interface 1682 It is necessary for the upstream CDN to have visibility into the 1683 delivery of content that it redirected to a downstream CDN. This 1684 allows the upstream CDN to properly bill its customers for multiple 1685 deliveries of content cached by the downstream CDN, as well as to 1686 report accurate traffic statistics to those content providers. This 1687 is one role of the LI. 1689 Other operational data that may be relevant to CDNI can also be 1690 exchanged by the LI. For example, dCDN may report the amount of 1691 content it has acquired from uCDN, and how much cache storage has 1692 been consumed by content cached on behalf of uCDN. 1694 Traffic logs are easily exchanged off-line. For example, the 1695 following traffic log is a small deviation from the Apache log file 1696 format, where entries include the following fields: 1698 o Domain - the full domain name of the origin server 1700 o IP address - the IP address of the client making the request 1702 o End time - the ending time of the transfer 1704 o Time zone - any time zone modifier for the end time 1706 o Method - the transfer command itself (e.g., GET, POST, HEAD) 1708 o URL - the requested URL 1710 o Version - the protocol version, such as HTTP/1.0 1712 o Response - a numeric response code indicating transfer result 1713 o Bytes Sent - the number of bytes in the body sent to the client 1715 o Request ID - a unique identifier for this transfer 1717 o User agent - the user agent, if supplied 1719 o Duration - the duration of the transfer in milliseconds 1721 o Cached Bytes - the number of body bytes served from the cache 1723 o Referrer - the referrer string from the client, if supplied 1725 Of these, only the Domain field is indirect in the downstream CDN--it 1726 is set to the CDN-Domain used by the upstream CDN rather than the 1727 actual origin server. This field could then used to filter traffic 1728 log entries so only those entries matching the upstream CDN are 1729 reported to the corresponding operator. Further discussion of the LI 1730 can be found in [I-D.ietf-cdni-logging]. 1732 One open question is who does the filtering. One option is that the 1733 downstream CDN filters its own logs, and passes the relevant records 1734 directly to each upstream peer. This requires that the downstream 1735 CDN knows the set of CDN-Domains that belong to each upstream peer. 1736 If this information is already exchanged between peers as part of 1737 another interface, then direct peer-to-peer reporting is 1738 straightforward. If it is not available, and operators do not wish 1739 to advertise the set of CDN-Domains they serve to their peers, then 1740 the second option is for each CDN to send both its non-local traffic 1741 records and the set of CDN-Domains it serves to an independent third- 1742 party (i.e., a CDN Exchange), which subsequently filters, merges, and 1743 distributes traffic records on behalf of each participating CDN 1744 operator. 1746 A second open question is how timely traffic information should be. 1747 For example, in addition to offline traffic logs, accurate real-time 1748 traffic monitoring might also be useful, but such information 1749 requires that the downstream CDN inform the upstream CDN each time it 1750 serves upstream content from its cache. The downstream CDN can do 1751 this, for example, by sending a conditional HTTP GET request (If- 1752 Modified-Since) to the upstream CDN each time it receives an HTTP GET 1753 request from one of its end-users. This allows the upstream CDN to 1754 record that a request has been issued for the purpose of real-time 1755 traffic monitoring. The upstream CDN can also use this information 1756 to validate the traffic logs received later from the downstream CDN. 1758 There is obviously a tradeoff between accuracy of such monitoring and 1759 the overhead of the downstream CDN having to go back to the upstream 1760 CDN for every request. 1762 Another design tradeoff in the LI is the degree of aggregation or 1763 summarization of data. One situation that lends itself to 1764 summarization is the delivery of HTTP adaptive streaming (HAS), since 1765 the large number of individual chunk requests potentially results in 1766 large volumes of logging information. This case is discussed below, 1767 but other forms of aggregation may also be useful. For example, 1768 there may be situations where bulk metrics such as bytes delivered 1769 per hour may suffice rather than the detailed per-request logs 1770 outlined above. It seems likely that a range of granularities of 1771 logging will be needed along with ways to specify the type and degree 1772 of aggregation required. 1774 4.5. CDNI Control Interface 1776 The CDNI Control interface is initially used to bootstrap the other 1777 interfaces. As a simple example, it could be used to provide the 1778 address of the logging server in dCDN to uCDN in order to bootstrap 1779 the CDNI Logging interface. It may also be used, for example, to 1780 establish security associations for the other interfaces. 1782 The other role the CI plays is to allow the uCDN to pre-position, 1783 revalidate, or purge metadata and content on a dCDN. These 1784 operations, sometimes collectively called the Trigger interface, are 1785 discussed further in [I-D.ietf-cdni-control-triggers]. 1787 4.6. CDNI Metadata Interface 1789 The role of the CDNI Metadata interface is to enable CDNI 1790 distribution metadata to be conveyed to the downstream CDN by the 1791 upstream CDN. Such metadata includes geo-blocking restrictions, 1792 availability windows, access control policies, and so on. It may 1793 also include information to facilitate acquisition of content by dCDN 1794 (e.g., alternate sources for the content, authorization information 1795 needed to acquire the content from the source). For a full 1796 discussion of the CDNI Metadata Interface, see 1797 [I-D.ietf-cdni-metadata] 1799 Some distribution metadata may be partially emulated using in-band 1800 mechanisms. For example, in case of any geo-blocking restrictions or 1801 availability windows, the upstream CDN can elect to redirect a 1802 request to the downstream CDN only if that CDN's advertised delivery 1803 footprint is acceptable for the requested URL. Similarly, the 1804 request could be forwarded only if the current time is within the 1805 availability window. However, such approaches typically come with 1806 shortcomings such as inability to prevent from replay outside the 1807 time window or inability to make use of a downstream CDN that covers 1808 a broader footprint than the geo-blocking restrictions. 1810 Similarly, some forms of access control may also be performed on a 1811 per-request basis using HTTP directives. For example, being able to 1812 respond to a conditional GET request gives the upstream CDN an 1813 opportunity to influence how the downstream CDN delivers its content. 1814 Minimally, the upstream CDN can invalidate (purge) content previously 1815 cached by the downstream CDN. 1817 Fine-grain control over how the downstream CDN delivers content on 1818 behalf of the upstream CDN is also possible. For example, by 1819 including the Forwarded HTTP header [I-D.ietf-appsawg-http-forwarded] 1820 with the conditional GET request, the downstream CDN can report the 1821 end-user's IP address to the upstream CDN, giving it an opportunity 1822 to control whether the downstream CDN should serve the content to 1823 this particular end-user. The upstream CDN would communicate its 1824 directive through its response to the conditional GET. The 1825 downstream CDN can cache information for a period of time specified 1826 by the upstream CDN, thereby reducing control overhead, but then 1827 preventing per-request control during the corresponding caching 1828 period. 1830 All of these in-band techniques serve to illustrate that uCDNs have 1831 the option of enforcing some of their access control policies 1832 themselves (at the expense of increased inter-CDN signaling load), 1833 rather than delegating enforcement to dCDNs using the MI. As a 1834 consequence, the MI could provide a means for the uCDN to express its 1835 desire to retain enforcement for itself. For example, this might be 1836 done by including a "check with me" flag in the metadata associated 1837 with certain content. The realization of such in-band techniques 1838 over the various inter-CDN acquisition protocols (e.g., HTTP) 1839 requires further investigation and may require small extensions or 1840 semantic changes to the acquisition protocol. 1842 4.7. HTTP Adaptive Streaming Concerns 1844 We consider HTTP Adaptive Streaming (HAS) and the impact it has on 1845 the CDNI interfaces because large objects (e.g., videos) are broken 1846 into a sequence of small, independent chunks. For each of the 1847 following, a more thorough discussion, including an overview of the 1848 tradeoffs involved in alternative designs, can be found in RFC 6983. 1850 First, with respect to Content Acquisition and File Management, which 1851 are out-of-scope for the CDNI interfaces but nontheless relevant to 1852 the overall operation, we assume no additional measures are required 1853 to deal with large numbers of chunks. This means that the dCDN is 1854 not explicitly made aware of any relationship between different 1855 chunks and the dCDN handles each chunk as if it were an individual 1856 and independent content item. The result is that content acquisition 1857 between uCDN and dCDN also happens on a per-chunk basis. This 1858 approach is in line with the recommendations made in RFC 6983, which 1859 also identifies potential improvements in this area that might be 1860 considered in the future. 1862 Second, with respect to Request Routing, we note that HAS manifest 1863 files have the potential to interfere with request routing since 1864 manifest files contain URLs pointing to the location of content 1865 chunks. To make sure that a manifest file does not hinder CDNI 1866 request routing and does not place excessive load on CDNI resources, 1867 the use of manifest files could either be limited to those containing 1868 relative URLs or the uCDN could modify the URLs in the manifest. Our 1869 approach for dealing with these issues is twofold. As a mandatory 1870 requirement, CDNs should be able to handle unmodified manifest files 1871 containing either relative or absolute URLs. To limit the number of 1872 redirects, and thus the load placed on the CDNI interfaces, as an 1873 optional feature uCDNs can use the information obtained through the 1874 CNDI Request Routing Redirection interface to modify the URLs in the 1875 manifest file. Since the modification of the manifest file is an 1876 optional uCDN-internal process, this does not require any 1877 standardization effort beyond being able to communicate chunk 1878 locations in the CDNI Request Routing Redirection interface. 1880 Third, with respect to the CDNI Logging interface, there are several 1881 potential issues, including the large number of individual chunk 1882 requests potentially resulting in large volumes of logging 1883 information, and the desire to correlate logging information for 1884 chunk requests that correspond to the same HAS session. For the 1885 initial CDNI specification, our approach is to expect participating 1886 CDNs to support per-chunk logging (e.g. logging each chunk request as 1887 if it were an independent content request) over the CDNI Logging 1888 interface. Optionally, the LI may include a Content Collection 1889 IDentifier (CCID) and/or a Session IDentifier (SID) as part of the 1890 logging fields, thereby facilitating correlation of per-chunk logs 1891 into per-session logs for applications benefiting from such session 1892 level information (e.g. session-based analytics). This approach is 1893 in line with the recommendations made in RFC 6983, which also 1894 identifies potential improvements in this area that might be 1895 considered in the future. 1897 Fourth, with respect to the CDNI Control interface, and in particular 1898 purging HAS chunks from a given CDN, our approach is to expect each 1899 CDN supports per-chunk content purge (e.g. purging of chunks as if 1900 they were individual content items). Optionally, a CDN may support 1901 content purge on the basis of a "Purge IDentifier (Purge-ID)" 1902 allowing the removal of all chunks related to a given Content 1903 Collection with a single reference. It is possible that this Purge- 1904 ID could be merged with the CCID discussed above for HAS Logging, or 1905 alternatively, they may remain distinct. 1907 4.8. URI Rewriting 1909 When using HTTP redirection, content URIs may be rewritten when 1910 redirection takes place within an uCDN, from an uCDN to a dCDN, and 1911 within the dCDN. In the case of cascaded CDNs, content URIs may be 1912 rewritten at every CDN hop (e.g., between the uCDN and the dCDN 1913 acting as the transit CDN, and between the transit CDN and the dCDN 1914 serving the request. The content URI used between any uCDN/dCDN pair 1915 becomes a common handle that can be referred without ambiguity by 1916 both CDNs in all their inter-CDN communications. This handle allows 1917 the uCDN and dCDN to correlate information exchanged using other CDNI 1918 interfaces in both the downstream direction (e.g., when using the MI) 1919 and the upstream direction (e.g., when using the LI). 1921 Consider the simple case of a single uCDN/dCDN pair using HTTP 1922 redirection. We introduce the following terminology for content URIs 1923 to simplify the discussion: 1925 "u-URI" represents a content URI in a request presented to the 1926 uCDN; 1928 "ud-URI" is a content URI acting as the common handle across uCDN 1929 and dCDN for requests redirected by the uCDN to a specific dCDN; 1931 "d-URI" represents a content URI in a request made within the 1932 delegate dCDN. 1934 In our simple pair-wise example, the "ud-URI" effectively becomes the 1935 handle that the uCDN/dCDN pair use to correlate all CDNI information. 1936 In particular, for a given pair of CDNs executing the HTTP 1937 redirection, the uCDN needs to map the u-URI to the ud-URI handle for 1938 all MI message exchanges, while the dCDN needs to map the d-URI to 1939 the ud-URI handle for all LI message exchanges. 1941 In the case of a cascaded CDNs, the transit CDN will rewrite the 1942 content URI when redirecting to the dCDN, thereby establishing a new 1943 handle between the transit CDN and the dCDN, that is different from 1944 the handle between the uCDN and transit CDN. It is the 1945 responsibility of the transit CDN to manage its mapping across 1946 handles so the right handle for all pairs of CDNs is always used in 1947 its CDNI communication. 1949 In summary, all CDNI interfaces between a given pair of CDNs need to 1950 always use the "ud-URI" handle for that specific CDN pair as their 1951 content URI reference. 1953 5. Deployment Models 1955 In this section we describe a number of possible deployment models 1956 that may be achieved using the CDNI interfaces described above. We 1957 note that these models are by no means exhaustive, and that many 1958 other models may be possible. 1960 Although the reference model of Figure 1 shows all CDN functions on 1961 each side of the CDNI interface, deployments can rely on entities 1962 that are involved in any subset of these functions, and therefore 1963 only support the relevant subset of CDNI interfaces. As already 1964 noted in Section 3, effective CDNI deployments can be built without 1965 necessarily implementing all the interfaces. Some examples of such 1966 deployments are shown below. 1968 Note that, while we refer to upstream and downstream CDNs, this 1969 distinction applies to specific content items and transactions. That 1970 is, a given CDN may be upstream for some transactions and downstream 1971 for others, depending on many factors such as location of the 1972 requesting client and the particular piece of content requested. 1974 5.1. Meshed CDNs 1976 Although the reference model illustrated in Figure 1 shows a 1977 unidirectional CDN interconnection with a single uCDN and a single 1978 dCDN, any arbitrary CDNI meshing can be built from this, such as the 1979 example meshing illustrated in Figure 11. (Support for arbitrary 1980 meshing may or may not be in the initial scope for the working group, 1981 but the model allows for it.) 1982 ------------- ----------- 1983 / CDN A \<==CDNI===>/ CDN B \ 1984 \ / \ / 1985 ------------- ----------- 1986 /\ \\ /\ 1987 || \\ || 1988 CDNI \==CDNI===\\ CDNI 1989 || \\ || 1990 \/ \/ \/ 1991 ------------- ----------- 1992 / CDN C \===CDNI===>/ CDN D \ 1993 \ / \ / 1994 ------------- ----------- 1995 /\ 1996 || 1997 CDNI 1998 || 1999 \/ 2000 ------------- 2001 / CDN E \ 2002 \ / 2003 ------------- 2005 ===> CDNI interfaces, with right-hand side CDN acting as dCDN 2006 to left-hand side CDN 2007 <==> CDNI interfaces, with right-hand side CDN acting as dCDN 2008 to left-hand side CDN and with left-hand side CDN acting 2009 as dCDN to right-hand side CDN 2011 Figure 11: CDNI Deployment Model: CDN Meshing Example 2013 5.2. CSP combined with CDN 2015 Note that our terminology refers to functional roles and not economic 2016 or business roles. That is, a given organization may be operating as 2017 both a CSP and a fully fledged uCDN when we consider the functions 2018 performed, as illustrated in Figure 12. 2020 ##################################### ################## 2021 # # # # 2022 # Organization A # # Organization B # 2023 # # # # 2024 # -------- ------------- # # ----------- # 2025 # / CSP \ / uCDN \ # # / dCDN \ # 2026 # | | | +----+ | # # | +----+ | # 2027 # | | | | C | | # # | | C | | # 2028 # | | | +----+ | # # | +----+ | # 2029 # | | | +----+ | # # | +----+ | # 2030 # | | | | L | | # # | | L | | # 2031 # | |*****| +----+ |===CDNI===>| +----+ | # 2032 # | | | +----+ | # # | +----+ | # 2033 # | | | | RR | | # # | | RR | | # 2034 # | | | +----+ | # # | +----+ | # 2035 # | | | +----+ | # # | +----+ | # 2036 # | | | | D | | # # | | D | | # 2037 # | | | +----+ | # # | +----+ | # 2038 # \ / \ / # # \ / # 2039 # -------- ------------- # # ----------- # 2040 # # # # 2041 ##################################### ################## 2043 ===> CDNI interfaces, with right-hand side CDN acting as dCDN 2044 to left-hand side CDN 2045 **** interfaces outside the scope of CDNI 2046 C Control component of the CDN 2047 L Logging component of the CDN 2048 RR Request Routing component of the CDN 2049 D Distribution component of the CDN 2051 Figure 12: CDNI Deployment Model: Organization combining CSP & uCDN 2053 5.3. CSP using CDNI Request Routing Interface 2055 As another example, a content provider organization may choose to run 2056 its own request routing function as a way to select among multiple 2057 candidate CDN providers; In this case the content provider may be 2058 modeled as the combination of a CSP and of a special, restricted case 2059 of a CDN. In that case, as illustrated in Figure 13, the CDNI 2060 Request Routing interfaces can be used between the restricted CDN 2061 operated by the content provider Organization and the CDN operated by 2062 the full CDN organization acting as a dCDN in the request routing 2063 control plane. Interfaces outside the scope of the CDNI work can be 2064 used between the CSP functional entities of the content provider 2065 organization and the CDN operated by the full CDN organization acting 2066 as a uCDN) in the CDNI control planes other than the request routing 2067 plane (i.e. Control, Distribution, Logging). 2069 ##################################### ################## 2070 # # # # 2071 # Organization A # # Organization B # 2072 # # # # 2073 # -------- ------------- # # ----------- # 2074 # / CSP \ / uCDN(RR) \ # # / dCDN(RR) \ # 2075 # | | | +----+ | # # | +----+ | # 2076 # | |*****| | RR |==========CDNI=====>| RR | | # 2077 # | | | +----+ | # RR # | +----+ | # 2078 # | | \ / # # | | # 2079 # | | ------------- # # |uCDN(C,L,D)| # 2080 # | | # # | +----+ | # 2081 # | | # # | | C | | # 2082 # | |*******************************| +----+ | # 2083 # | | # # | +----+ | # 2084 # | | # # | | L | | # 2085 # | | # # | +----+ | # 2086 # | | # # | +----+ | # 2087 # | | # # | | D | | # 2088 # | | # # | +----+ | # 2089 # \ / # # \ / # 2090 # -------- # # ----------- # 2091 # # # # 2092 ##################################### ################## 2094 ===> CDNI Request Routing Interface 2095 **** interfaces outside the scope of CDNI 2097 Figure 13: CDNI Deployment Model: Organization combining CSP and 2098 partial CDN 2100 5.4. CDN Federations and CDN Exchanges 2102 There are two additional concepts related to, but distinct from CDN 2103 Interconnection. The first is CDN Federation. Our view is that CDNI 2104 is the more general concept, involving two or more CDNs serving 2105 content to each other's users, while federation implies a multi- 2106 lateral interconnection arrangement, but other CDN interconnection 2107 agreements are also possible (e.g., symmetric bilateral, asymmetric 2108 bilateral). An important conclusion is that CDNI technology should 2109 not presume (or bake in) a particular interconnection agreement, but 2110 should instead be general enough to permit alternative 2111 interconnection arrangements to evolve. 2113 The second concept often used in the context of CDN Federation is CDN 2114 Exchange--a third party broker or exchange that is used to facilitate 2115 a CDN federation. Our view is that a CDN exchange offers valuable 2116 machinery to scale the number of CDN operators involved in a multi- 2117 lateral (federated) agreement, but that this machinery is built on 2118 top of the core CDNI interconnection mechanisms. For example, as 2119 illustrated in Figure 14, the exchange might aggregate and 2120 redistribute information about each CDN footprint and capacity, as 2121 well as collect, filter, and redistribute traffic logs that each 2122 participant needs for interconnection settlement, but inter-CDN 2123 request routing, inter-CDN content distribution (including inter-CDN 2124 acquisition) and inter-CDN control which fundamentally involve a 2125 direct interaction between an upstream CDN and a downstream CDN-- 2126 operate exactly as in a pair-wise peering arrangement. Turning to 2127 Figure 14, we observe that in this example: 2129 o each CDN supports a direct CDNI Control interface to every other 2130 CDN 2132 o each CDN supports a direct CDNI Metadata interface to every other 2133 CDN 2135 o each CDN supports a CDNI Logging interface with the CDN Exchange 2137 o each CDN supports both a CDNI Request Routing interfaces with the 2138 CDN Exchange (for aggregation and redistribution of dynamic CDN 2139 footprint discovery information) and a direct RI to every other 2140 CDN (for actual request redirection). 2142 ---------- --------- 2143 / CDN A \ / CDN B \ 2144 | +----+ | | +----+ | 2145 //========>| C |<==============CDNI============>| C |<==========\\ 2146 || | +----+ | C | +----+ | || 2147 || | +----+ | | +----+ | || 2148 || //=====>| D |<==============CDNI============>| D |<=======\\ || 2149 || || | +----+ | M | +----+ | || || 2150 || || | | /------------\ | | || || 2151 || || | +----+ | | +--+ CDN Ex| | +----+ | || || 2152 || || //==>| RR |<===CDNI==>|RR|<=======CDNI====>| RR |<====\\ || || 2153 || || || | +----+ | RR | +--+ | RR | +----+ | || || || 2154 || || || | | | /\ | | | || || || 2155 || || || | +----+ | | || +---+ | | +----+ | || || || 2156 || || || | | L |<===CDNI=======>| L |<=CDNI====>| L | | || || || 2157 || || || | +----+ | L | || +---+ | L | +----+ | || || || 2158 || || || \ / \ || /\ / \ / || || || 2159 || || || ----------- --||----||-- ----------- || || || 2160 || || || || || || || || 2161 || || || CDNI RR || || || || 2162 || || || || CDNI L || || || 2163 || || || || || || || || 2164 || || || ---||----||---- || || || 2165 || || || / \/ || \ || || || 2166 || || || | +----+ || | || || || 2167 || || \\=====CDNI==========>| RR |<=============CDNI========// || || 2168 || || RR | +----+ \/ | RR || || 2169 || || | +----+ | || || 2170 || || | | L | | || || 2171 || || | +----+ | || || 2172 || || | +----+ | || || 2173 || \\=======CDNI===========>| D |<=============CDNI===========// || 2174 || M | +----+ | M || 2175 || | +----+ | || 2176 \\==========CDNI===========>| C |<=============CDNI==============// 2177 C | +----+ | C 2178 \ CDN C / 2179 -------------- 2181 <=CDNI RR=> CDNI Request Routing Interface 2182 <=CDNI M==> CDNI Metadata Interface 2183 <=CDNI C==> CDNI Control Interface 2184 <=CDNI L==> CDNI Logging Interface 2186 Figure 14: CDNI Deployment Model: CDN Exchange 2188 Note that a CDN exchange may alternatively support a different set of 2189 functionality (e.g. Logging only, or Logging and full request 2190 routing, or all the functionality of a CDN including content 2191 distribution). All these options are expected to be allowed by the 2192 IETF CDNI specifications. 2194 6. Trust Model 2196 There are a number of trust issues that need to be addressed by a 2197 CDNI solution. Many of them are in fact similar or identical to 2198 those in a simple CDN without interconnection. In a standard CDN 2199 environment (without CDNI), the CSP places a degree of trust in a 2200 single CDN operator to perform many functions. The CDN is trusted to 2201 deliver content with appropriate quality of experience for the end 2202 user. The CSP trusts the CDN operator not to corrupt or modify the 2203 content. The CSP often relies on the CDN operator to provide 2204 reliable accounting information regarding the volume of delivered 2205 content. The CSP may also trust the CDN operator to perform actions 2206 such as timely invalidation of content and restriction of access to 2207 content based on certain criteria such as location of the user and 2208 time of day, and to enforce per-request authorization performed by 2209 the CSP using techniques such as URI signing. 2211 A CSP also places trust in the CDN not to distribute any information 2212 that is confidential to the CSP (e.g., how popular a given piece of 2213 content is) or confidential to the end user (e.g., which content has 2214 been watched by which user). 2216 A CSP does not necessarily have to place complete trust in a CDN. A 2217 CSP will in some cases take steps to protect its content from 2218 improper distribution by a CDN, e.g. by encrypting it and 2219 distributing keys in some out of band way. A CSP also depends on 2220 monitoring (possibly by third parties) and reporting to verify that 2221 the CDN has performed adequately. A CSP may use techniques such as 2222 client-based metering to verify that accounting information provided 2223 by the CDN is reliable. HTTP conditional requests may be used to 2224 provide the CSP with some checks on CDN operation. In other words, 2225 while a CSP may trust a CDN to perform some functions in the short 2226 term, the CSP is able in most cases to verify whether these actions 2227 have been performed correctly and to take action (such as moving the 2228 content to a different CDN) if the CDN does not live up to 2229 expectations. 2231 The main trust issue raised by CDNI is that it introduces transitive 2232 trust. A CDN that has a direct relationship with a CSP can now 2233 "outsource" the delivery of content to another (downstream) CDN. 2234 That CDN may in term outsource delivery to yet another downstream 2235 CDN, and so on. 2237 The top level CDN in such a chain of delegation is responsible for 2238 ensuring that the requirements of the CSP are met. Failure to do so 2239 is presumably just as serious as in the traditional single CDN case. 2240 Hence, an upstream CDN is essentially trusting a downstream CDN to 2241 perform functions on its behalf in just the same way as a CSP trusts 2242 a single CDN. Monitoring and reporting can similarly be used to 2243 verify that the downstream CDN has performed appropriately. However, 2244 the introduction of multiple CDNs in the path between CSP and end 2245 user complicates the picture. For example, third party monitoring of 2246 CDN performance (or other aspects of operation, such as timely 2247 invalidation) might be able to identify the fact that a problem 2248 occurred somewhere in the chain but not point to the particular CDN 2249 at fault. 2251 In summary, we assume that an upstream CDN will invest a certain 2252 amount of trust in a downstream CDN, but that it will verify that the 2253 downstream CDN is performing correctly, and take corrective action 2254 (including potentially breaking off its relationship with that CDN) 2255 if behavior is not correct. We do not expect that the trust 2256 relationship between a CSP and its "top level" CDN will differ 2257 significantly from that found today in single CDN situations. 2258 However, it does appear that more sophisticated tools and techniques 2259 for monitoring CDN performance and behavior will be required to 2260 enable the identification of the CDN at fault in a particular 2261 delivery chain. 2263 We expect that the detailed designs for the specific interfaces for 2264 CDNI will need to take the transitive trust issues into account. For 2265 example, explicit confirmation that some action (such as content 2266 removal) has taken place in a downstream CDN may help to mitigate 2267 some issues of transitive trust. 2269 7. IANA Considerations 2271 This memo includes no request to IANA. 2273 8. Security Considerations 2275 While there is a variety of security issues introduced by a single 2276 CDN, we are concerned here specifically with the additional issues 2277 that arise when CDNs are interconnected. For example, when a single 2278 CDN has the ability to distribute content on behalf of a CSP, there 2279 may be concerns that such content could be distributed to parties who 2280 are not authorized to receive it, and there are mechanisms to deal 2281 with such concerns. Our focus in this section is on how CDN 2282 interconnection introduces new security issues not found in the 2283 single CDN case. 2285 Many of the security issues that arise in CDNI are related to the 2286 transitivity of trust (or lack thereof) described in Section 6. As 2287 noted above, the design of the various interfaces for CDNI must take 2288 account of the additional risks posed by the fact that a CDN with 2289 whom a CSP has no direct relationship is now potentially distributing 2290 content for that CSP. The mechanisms used to mitigate these risks 2291 may be similar to those used in the single CDN case, but their 2292 suitability in this more complex environment must be validated. 2294 Another concern that arises in any CDN is that information about the 2295 behavior of users (what content they access, how much content they 2296 consume, etc.) may be gathered by the CDN. This risk certainly 2297 exists in inter-connected CDNs, but it should be possible to apply 2298 the same techniques to mitigate it as in the single CDN case. 2300 CDNs today offer a variety of means to control access to content, 2301 such as time-of-day restrictions, geo-blocking, and URI signing. 2302 These mechanisms must continue to function in CDNI environments, and 2303 this consideration is likely to affect the design of certain CDNI 2304 interfaces (e.g. metadata, request routing.) 2306 Just as with a single CDN, each peer CDN must ensure that it is not 2307 used as an "open proxy" to deliver content on behalf of a malicious 2308 CSP. Whereas a single CDN typically addresses this problem by having 2309 CSPs explicitly register content (or origin servers) that is to be 2310 served, simply propagating this information to peer downstream CDNs 2311 may be problematic because it reveals more information than the 2312 upstream CDN is willing to specify. (To this end, the content 2313 acquisition step in the earlier examples force the dCDN to retrieve 2314 content from the uCDN rather than go directly to the origin server.) 2316 There are several approaches to this problem. One is for the uCDN to 2317 encoded a signed token generated from a shared secret in each URL 2318 routed to a dCDN, and for the dCDN to validate the request based on 2319 this token. Another one is to have each upstream CDN advertise the 2320 set of CDN-Domains they serve, where the downstream CDN checks each 2321 request against this set before caching and delivering the associated 2322 object. Although straightforward, this approach requires operators 2323 to reveal additional information, which may or may not be an issue. 2325 8.1. Security of CDNI Interfaces 2327 It is noted in [I-D.ietf-cdni-requirements] that all CDNI interfaces 2328 must be able to operate securely over insecure IP networks. Since it 2329 is expected that the CDNI interfaces will be implemented using 2330 existing application protocols such as HTTP or XMPP, we also expect 2331 that the security mechanisms available to those protocols may be used 2332 by the CDNI interfaces. Details of how these interfaces are secured 2333 will be specified in the relevant interface documents. 2335 8.2. Digital Rights Management 2337 Issues of digital rights management (DRM, also sometimes called 2338 digital restrictions management) is often employed for content 2339 distributed via CDNs. In general, DRM relies on the CDN to 2340 distribute encrypted content, with decryption keys distributed to 2341 users by some other means (e.g. directly from the CSP to the end 2342 user.) For this reason, DRM is considered out of scope [RFC6707] and 2343 does not introduce additional security issues for CDNI. 2345 9. Contributors 2347 The following individuals contributed to this document: 2349 o Ray van Brandenburg 2351 o Matt Caulfield 2353 o Francois le Faucheur 2355 o Aaron Falk 2357 o David Ferguson 2359 o John Hartman 2361 o Ben Niven-Jenkins 2363 o Kent Leung 2365 10. Acknowledgements 2367 We thank Huw Jones for helpful input to the draft. 2369 11. Informative References 2371 [I-D.ietf-appsawg-http-forwarded] 2372 Petersson, A. and M. Nilsson, "Forwarded HTTP Extension", 2373 draft-ietf-appsawg-http-forwarded-10 (work in progress), 2374 October 2012. 2376 [I-D.ietf-cdni-control-triggers] 2377 Murray, R. and B. Niven-Jenkins, "CDNI Control Interface / 2378 Triggers", draft-ietf-cdni-control-triggers-02 (work in 2379 progress), December 2013. 2381 [I-D.ietf-cdni-footprint-capabilities-semantics] 2382 Seedorf, J., Peterson, J., Previdi, S., Brandenburg, R., 2383 and K. Ma, "CDNI Request Routing: Footprint and 2384 Capabilities Semantics", draft-ietf-cdni-footprint- 2385 capabilities-semantics-01 (work in progress), October 2386 2013. 2388 [I-D.ietf-cdni-logging] 2389 Faucheur, F., Bertrand, G., Oprescu, I., and R. 2390 Peterkofsky, "CDNI Logging Interface", draft-ietf-cdni- 2391 logging-08 (work in progress), October 2013. 2393 [I-D.ietf-cdni-metadata] 2394 Niven-Jenkins, B., Murray, R., Watson, G., Caulfield, M., 2395 Leung, K., and K. Ma, "CDN Interconnect Metadata", draft- 2396 ietf-cdni-metadata-04 (work in progress), December 2013. 2398 [I-D.ietf-cdni-redirection] 2399 Danhua, W., Niven-Jenkins, B., He, X., Chen, G., and W. 2400 Ni, "Request Routing Redirection Interface for CDN 2401 Interconnection", draft-ietf-cdni-redirection-01 (work in 2402 progress), October 2013. 2404 [I-D.ietf-cdni-requirements] 2405 Leung, K. and Y. Lee, "Content Distribution Network 2406 Interconnection (CDNI) Requirements", draft-ietf-cdni- 2407 requirements-16 (work in progress), January 2014. 2409 [I-D.vandergaast-edns-client-subnet] 2410 Contavalli, C., Gaast, W., Leach, S., and E. Lewis, 2411 "Client Subnet in DNS Requests", draft-vandergaast-edns- 2412 client-subnet-02 (work in progress), July 2013. 2414 [RFC3466] Day, M., Cain, B., Tomlinson, G., and P. Rzewski, "A Model 2415 for Content Internetworking (CDI)", RFC 3466, February 2416 2003. 2418 [RFC6707] Niven-Jenkins, B., Le Faucheur, F., and N. Bitar, "Content 2419 Distribution Network Interconnection (CDNI) Problem 2420 Statement", RFC 6707, September 2012. 2422 [RFC6770] Bertrand, G., Stephan, E., Burbridge, T., Eardley, P., Ma, 2423 K., and G. Watson, "Use Cases for Content Delivery Network 2424 Interconnection", RFC 6770, November 2012. 2426 [RFC6983] van Brandenburg, R., van Deventer, O., Le Faucheur, F., 2427 and K. Leung, "Models for HTTP-Adaptive-Streaming-Aware 2428 Content Distribution Network Interconnection (CDNI)", RFC 2429 6983, July 2013. 2431 Authors' Addresses 2433 Larry Peterson 2434 Akamai Technologies, Inc. 2435 8 Cambridge Center 2436 Cambridge, MA 02142 2437 USA 2439 Email: lapeters@akamai.com 2441 Bruce Davie 2442 VMware, Inc. 2443 3401 Hillview Ave. 2444 Palo Alto, CA 94304 2445 USA 2447 Email: bdavie@vmware.com 2449 Ray van Brandenburg (editor) 2450 TNO 2451 Brassersplein 2 2452 Delft 2612CT 2453 the Netherlands 2455 Phone: +31-88-866-7000 2456 Email: ray.vanbrandenburg@tno.nl