idnits 2.17.1 draft-ietf-cdni-framework-08.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 : ---------------------------------------------------------------------------- == There are 2 instances of lines with non-RFC2606-compliant FQDNs in the document. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (January 19, 2014) is 3744 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 ---------------------------------------------------------------------------- == Missing Reference: 'RI REPLY' is mentioned on line 474, but not defined == Outdated reference: A later version (-15) exists of draft-ietf-cdni-control-triggers-02 == Outdated reference: A later version (-20) exists of draft-ietf-cdni-footprint-capabilities-semantics-01 == Outdated reference: A later version (-27) exists of draft-ietf-cdni-logging-08 == Outdated reference: A later version (-21) exists of draft-ietf-cdni-metadata-04 == Outdated reference: A later version (-20) exists of draft-ietf-cdni-redirection-01 == Outdated reference: A later version (-17) exists of draft-ietf-cdni-requirements-14 -- Obsolete informational reference (is this intentional?): RFC 3466 (Obsoleted by RFC 7336) Summary: 0 errors (**), 0 flaws (~~), 9 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group L. Peterson, Ed. 3 Internet-Draft Akamai Technologies, Inc. 4 Obsoletes: 3466 (if approved) B. Davie 5 Intended status: Informational VMware, Inc. 6 Expires: July 23, 2014 January 19, 2014 8 Framework for CDN Interconnection 9 draft-ietf-cdni-framework-08 11 Abstract 13 This document presents a framework for Content Distribution Network 14 Interconnection (CDNI). The purpose of the framework is to provide 15 an overall picture of the problem space of CDNI and to describe the 16 relationships among the various components necessary to interconnect 17 CDNs. CDN Interconnection requires the specification of interfaces 18 and mechanisms to address issues such as request routing, 19 distribution metadata exchange, and logging information exchange 20 across CDNs. The intent of this document is to outline what each 21 interface needs to accomplish, and to describe how these interfaces 22 and mechanisms fit together, while leaving their detailed 23 specification to other documents. It obsoletes RFC 3466. 25 Status of This Memo 27 This Internet-Draft is submitted in full conformance with the 28 provisions of BCP 78 and BCP 79. 30 Internet-Drafts are working documents of the Internet Engineering 31 Task Force (IETF). Note that other groups may also distribute 32 working documents as Internet-Drafts. The list of current Internet- 33 Drafts is at http://datatracker.ietf.org/drafts/current/. 35 Internet-Drafts are draft documents valid for a maximum of six months 36 and may be updated, replaced, or obsoleted by other documents at any 37 time. It is inappropriate to use Internet-Drafts as reference 38 material or to cite them other than as "work in progress." 40 This Internet-Draft will expire on July 23, 2014. 42 Copyright Notice 44 Copyright (c) 2014 IETF Trust and the persons identified as the 45 document authors. All rights reserved. 47 This document is subject to BCP 78 and the IETF Trust's Legal 48 Provisions Relating to IETF Documents 49 (http://trustee.ietf.org/license-info) in effect on the date of 50 publication of this document. Please review these documents 51 carefully, as they describe your rights and restrictions with respect 52 to this document. Code Components extracted from this document must 53 include Simplified BSD License text as described in Section 4.e of 54 the Trust Legal Provisions and are provided without warranty as 55 described in the Simplified BSD License. 57 Table of Contents 59 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 60 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . . . 3 61 1.2. Reference Model . . . . . . . . . . . . . . . . . . . . . 4 62 1.3. Structure Of This Document . . . . . . . . . . . . . . . 8 63 2. Building Blocks . . . . . . . . . . . . . . . . . . . . . . . 8 64 2.1. Request Redirection . . . . . . . . . . . . . . . . . . . 8 65 2.1.1. DNS Redirection . . . . . . . . . . . . . . . . . . . 8 66 2.1.2. HTTP Redirection . . . . . . . . . . . . . . . . . . 9 67 3. Overview of CDNI Operation . . . . . . . . . . . . . . . . . 10 68 3.1. Preliminaries . . . . . . . . . . . . . . . . . . . . . . 12 69 3.2. Iterative HTTP Redirect Example . . . . . . . . . . . . . 13 70 3.3. Recursive HTTP Redirection Example . . . . . . . . . . . 18 71 3.4. Iterative DNS-based Redirection Example . . . . . . 22 72 3.5. Dynamic Footprint Discovery Example . . . . . . . . . . . 25 73 3.6. Content Removal Example . . . . . . . . . . . . . . . . . 27 74 3.7. Pre-Positioned Content Acquisition Example . . . . . . . 28 75 3.8. Asynchronous CDNI Metadata Example . . . . . . . . . . . 29 76 3.9. Synchronous CDNI Metadata Acquisition Example . . . . . . 31 77 3.10. Content and Metadata Acquisition with Multiple Upstream 78 CDNs . . . . . . . . . . . . . . . . . . . . . . . . . . 33 79 4. Main Interfaces . . . . . . . . . . . . . . . . . . . . . . . 34 80 4.1. In-Band versus Out-of-Band Interfaces . . . . . . . . . . 35 81 4.2. Cross Interface Concerns . . . . . . . . . . . . . . . . 35 82 4.3. Request Routing Interfaces . . . . . . . . . . . . . . . 36 83 4.4. CDNI Logging Interface . . . . . . . . . . . . . . . . . 37 84 4.5. CDNI Control Interface . . . . . . . . . . . . . . . . . 39 85 4.6. CDNI Metadata Interface . . . . . . . . . . . . . . . . . 39 86 4.7. HTTP Adaptive Streaming Concerns . . . . . . . . . . . . 40 87 4.8. URI Rewriting . . . . . . . . . . . . . . . . . . . . . . 42 88 5. Deployment Models . . . . . . . . . . . . . . . . . . . . . . 43 89 5.1. Meshed CDNs . . . . . . . . . . . . . . . . . . . . . . . 43 90 5.2. CSP combined with CDN . . . . . . . . . . . . . . . . . . 44 91 5.3. CSP using CDNI Request Routing Interface . . . . . . . . 45 92 5.4. CDN Federations and CDN Exchanges . . . . . . . . . . . . 46 93 6. Trust Model . . . . . . . . . . . . . . . . . . . . . . . . . 49 94 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 50 95 8. Security Considerations . . . . . . . . . . . . . . . . . . . 50 96 8.1. Security of CDNI Interfaces . . . . . . . . . . . . . . . 51 97 8.2. Digital Rights Management . . . . . . . . . . . . . . . . 52 98 9. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 52 99 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 52 100 11. Informative References . . . . . . . . . . . . . . . . . . . 52 101 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 54 103 1. Introduction 105 This document provides an overview of the various components 106 necessary to interconnect CDNs, expanding on the problem statement 107 and use cases introduced in [RFC6770] and [RFC6707]. It describes 108 the necessary interfaces and mechanisms in general terms and outlines 109 how they fit together to form a complete system for CDN 110 Interconnection. Detailed specifications are left to other 111 documents. This document makes extensive use of message flow 112 examples to illustrate the operation of interconnected CDNs, but 113 these examples should be considered illustrative rather than 114 prescriptive. 116 [RFC3466] uses different terminology and models for "Content 117 Internetworking (CDI)". It is also less prescriptive in terms of 118 interfaces. To avoid confusion, this document obsoletes [RFC3466]. 120 1.1. Terminology 122 This document uses the core terminology defined in [RFC6707]. It 123 also introduces the following terms: 125 CDN-Domain: a host name (FQDN) at the beginning of a URL, 126 representing a set of content that is served by a given CDN. For 127 example, in the URL http://cdn.csp.example/...rest of url..., the CDN 128 domain is cdn.csp.example. A major role of CDN-Domain is to identify 129 a region (subset) of the URI space relative to which various CDN 130 Interconnection rules and policies are to apply. For example, a 131 record of CDN Metadata might be defined for the set of resources 132 corresponding to some CDN-Domain. 134 Distinguished CDN-Domain: a CDN-Domain that is allocated by a CDN for 135 the purposes of communication with a peer CDN, but which is not found 136 in client requests. Such CDN-Domains may be used for inter-CDN 137 acquisition, or as redirection targets, and enable a CDN to 138 distinguish a request from a peer CDN from an end-user request. 140 Delivering CDN: the CDN that ultimately delivers a piece of content 141 to the end-user. The last in a potential sequence of downstream 142 CDNs. 144 Recursive CDNI Request Redirection: When an upstream CDN elects to 145 redirect a request towards a downstream CDN, the upstream CDN can 146 query the downstream CDN Request Routing system via the CDNI Request 147 Routing Redirection Interface (or use information cached from earlier 148 similar queries) to find out how the downstream CDN wants the request 149 to be redirected, which allows the upstream CDN to factor in the 150 downstream CDN response when redirecting the user agent. This 151 approach is referred to as "Recursive" CDNI Request Redirection. 152 Note that the downstream CDN may elect to have the request redirected 153 directly to a Surrogate inside the downstream CDN, to the Request 154 Routing System of the downstream CDN, to another CDN, or to whatever 155 system is necessary to handle the redirected request appropriately. 157 Iterative CDNI Request Redirection: When an upstream CDN elects to 158 redirect a request towards a downstream CDN, the upstream CDN can 159 base its redirection purely on a local decision (and without 160 attempting to take into account how the downstream CDN may in turn 161 redirect the user agent). In that case, the upstream CDN redirects 162 the request to the request routing system in the downstream CDN, 163 which in turn will decide how to redirect that request: this approach 164 is referred to as "Iterative" CDNI Request Redirection. 166 Synchronous CDNI operations: operations between CDNs that happen 167 during the process of servicing a user request, i.e. between the time 168 that the user agent begins its attempt to obtain content and the time 169 at which that request is served. 171 Asynchronous CDNI operations: operations between CDNs that happen 172 independently of any given user request, such as advertisement of 173 footprint information or pre-positioning of content for later 174 delivery. 176 Trigger Interface: a subset of the CDNI Control interface that 177 includes operations to pre-position, revalidate, and purge both 178 metadata and content. These operations are typically called in 179 response to some action (Trigger) by the CSP on the upstream CDN. 181 We also sometimes use uCDN and dCDN as shorthand for upstream CDN and 182 downstream CDN, respectively. 184 1.2. Reference Model 186 This document uses the reference model in Figure 1, which expands the 187 reference model originally defined in [RFC6707]. (The difference is 188 that the expanded model splits the Request Routing Interface into its 189 two distinct parts: the Request Routing Redirection interface and the 190 Footprint and Capabilities Advertisement interface, as described 191 below.) 192 -------- 193 / \ 194 | CSP | 195 \ / 196 -------- 197 * 198 * 199 * /\ 200 * / \ 201 ---------------------- |CDNI| ---------------------- 202 / Upstream CDN \ | | / Downstream CDN \ 203 | +-------------+ | | CI | | +-------------+ | 204 |******* Control |<======|====|=======>| Control *******| 205 |* +------*----*-+ | | | | +-*----*------+ *| 206 |* * * | | | | * * *| 207 |* +------*------+ | | LI | | +------*------+ *| 208 |* ***** Logging |<======|====|=======>| Logging ***** *| 209 |* * +-*-----------+ | | | | +-----------*-+ * *| 210 |* * * * | | | | * * * *| 211 .....*...+-*---------*-+ | | RI | | +-*---------*-+...*.*... 212 . |* * | |<======|====|=======>| | * *| . 213 . |* * | Req-Routing | | |FCI | | | Req-Routing | * *| . 214 . |* * *** |<======|====|=======>| |** * *| . 215 . |* * * +-------------+.| | | | +-------------+ * * *| . 216 . |* * * . | | | * * *| . 217 . |* * * +-------------+ |. | MI | | +-------------+ * * *| . 218 . |* * * | Distribution|<==.===|====|=======>| Distribution| * * *| . 219 . |* * * | | | . \ / | | | * * *| . 220 . |* * * |+---------+ | | . \/ | | +---------+| * * *| . 221 . |* * ***| +---------+| | ...Request......+---------+ |*** * *| . 222 . |* *****+-|Surrogate|***********************|Surrogate|-+***** *| . 223 . |******* +---------+| | Acquisition | |+----------+ *******| . 224 . | +-------------+ | | +-------*-----+ | . 225 . \ / \ * / . 226 . ---------------------- ---------*------------ . 227 . * . 228 . * Delivery . 229 . * . 230 . +--*---+ . 231 ...............Request............................| User |..Request.. 232 | Agent| 233 +------+ 235 <==> interfaces inside the scope of CDNI 237 **** and .... interfaces outside the scope of CDNI 239 Figure 1: CDNI Expanded Model and CDNI Interfaces 241 We note that while some interfaces in the reference model are "out of 242 scope" for the CDNI WG (in the sense that there is no need to define 243 new protocols for those interfaces) we still need to refer to them in 244 this document to explain the overall operation of CDNI. 246 We also note that, while we generally show only one upstream CDN 247 serving a given CSP, it is entirely possible that multiple uCDNs can 248 serve a single CSP. In fact, this situation effectively exists today 249 in the sense that a single CSP can currently delegate its content 250 delivery to more than one CDN. 252 The following briefly describes the five CDNI interfaces, 253 paraphrasing the definitions given in [RFC6707]. We discuss these 254 interfaces in more detail in Section 4. 256 o CDNI Control interface (CI): Operations to bootstrap and 257 parameterize the other CDNI interfaces, as well as operations to 258 pre-position, revalidate, and purge both metadata and content. 259 The latter subset of operations is sometimes collectively called 260 the "Trigger interface." 262 o CDNI Request Routing interface: Operations to determine what CDN 263 (and optionally what surrogate within a CDN) is to serve end- 264 user's requests. Is actually a logical bundling of two separate 265 but related interfaces: 267 * CDNI Footprint & Capabilities Advertisement interface (FCI): 268 Asynchronous operations to exchange routing information (e.g., 269 the network footprint and capabilities served by a given CDN) 270 that enables CDN selection for subsequent user requests; and 272 * CDNI Request Routing Redirection interface (RI): Synchronous 273 operations to select a delivery CDN (surrogate) for a given 274 user request. 276 o CDNI Metadata interface (MI): Operations to communicate metadata 277 that governs how the content is delivered by interconnected CDNs. 278 Examples of CDNI metadata include geo-blocking directives, 279 availability windows, access control mechanisms, and purge 280 directives. It may include a combination of: 282 * Asynchronous operations to exchange metadata that govern 283 subsequent user requests for content; and 285 * Synchronous operations that govern behavior for a given user 286 request for content. 288 o CDNI Logging interface (LI): Operations that allow interconnected 289 CDNs to exchange relevant activity logs. May include a 290 combination of: 292 * Real-time exchanges, suitable for runtime traffic monitoring; 293 and 295 * Offline exchanges, suitable for analytics and billing. 297 There is some potential overlap between the set of Trigger-based 298 operations in the CDNI Control interface and the CDNI Metadata 299 interface. For both cases, the information passed from the upstream 300 CDN to the downstream CDN can broadly be viewed as metadata that 301 describes how content is to be managed by the downstream CDN. For 302 example, the information conveyed by CI to pre-position, revalidate 303 or purge metadata is similar to the information conveyed by posting 304 updated metadata via the MI. Even the CI operation to purge content 305 could be viewed as an metadata update for that content: purge simply 306 says that the availability window for the named content ends now. 307 The two interfaces share much in common, so minimally, there will 308 need to be a consistent data model that spans both. 310 The distinction we draw has to do with what the uCDN knows about the 311 successful application of the metadata by the dCDN. In the case of 312 the CI, the downstream CDN returning a successful status message 313 guarantees that the operation has been successfully completed; e.g., 314 the content has been purged or pre-positioned. This implies that the 315 downstream CDN accepts responsibility for having successfully 316 completed the requested operation. In contrast, metadata passed 317 between CDNs via the MI carries no such completion guarantee. 318 Returning success implies successful receipt of the metadata, but 319 nothing can be inferred about precisely when the metadata will take 320 effect in the downstream CDN, only that it will take effect 321 eventually. This is because of the challenge in globally 322 synchronizing updates to metadata with end-user requests that are 323 currently in progress (or indistinguishable from currently being in 324 progress). Clearly, a CDN will not be viewed as a trusted peer if 325 "eventually" often becomes an indefinite period of time, but the 326 acceptance of responsibility cannot be as crisply defined for the MI. 328 Finally, there is a practical issue that impacts all of the CNDI 329 interfaces, and that is whether or not to optimize CDNI for HTTP 330 Adaptive Streaming (HAS). We highlight specific issues related to 331 delivering HAS content throughout this document, but for a more 332 thorough treatment of the topic, see [RFC6983]. 334 1.3. Structure Of This Document 336 The remainder of this document is organized as follows: 338 o Section 2 describes some essential building blocks for CDNI, 339 notably the various options for redirecting user requests to a 340 given CDN. 342 o Section 3 provides a number of illustrative examples of various 343 CDNI operations. 345 o Section 4 describes the functionality of the main CDNI interfaces. 347 o Section 5 shows how various deployment models of CDNI may be 348 achieved using the defined interfaces. 350 o Section 6 describes the trust model of CDNI and the issues of 351 transitive trust in particular that CDNI raises. 353 2. Building Blocks 355 2.1. Request Redirection 357 At its core, CDN Interconnection requires the redirection of requests 358 from one CDN to another. For any given request that is received by 359 an upstream CDN, it will either respond to the request directly, or 360 somehow redirect the request to a downstream CDN. Two main 361 mechanisms are available for redirecting a request to a downstream 362 CDN. The first leverages the DNS name resolution process and the 363 second uses in-protocol redirection mechanisms such as the HTTP 302 364 or 307 redirection response. We discuss these below as background 365 before discussing some examples of their use in Section 3. 367 2.1.1. DNS Redirection 369 DNS redirection is based on returning different IP addresses for the 370 same DNS name, for example, to balance server load or to account for 371 the client's location in the network. A DNS server, sometimes called 372 the Local DNS (LDNS), resolves DNS names on behalf of an end-user. 373 The LDNS server in turn queries other DNS servers until it reaches 374 the authoritative DNS server for the CDN-Domain. The network 375 operator typically provides the LDNS server, although the user is 376 free to choose other DNS servers (e.g., OpenDNS, Google Public DNS). 377 This latter possibility is important because the authoritative DNS 378 server sees only the IP address of the DNS server that queries it, 379 not the IP address of the original end-user. 381 The advantage of DNS redirection is that it is completely transparent 382 to the end user; the user sends a DNS name to the LDNS server and 383 gets back an IP address. On the other hand, DNS redirection is 384 problematic because the DNS request comes from the LDNS server, not 385 the end-user. This may affect the accuracy of server selection that 386 is based on the user's location. The transparency of DNS redirection 387 is also a problem in that there is no opportunity to take the 388 attributes of the user agent or the URI path component into account. 389 We consider two main forms of DNS redirection: simple and CNAME- 390 based. 392 In simple DNS redirection, the authoritative DNS server for the name 393 simply returns an IP address from a set of possible IP addresses. 394 The answer is chosen from the set based on characteristics of the set 395 (e.g., the relative loads on the servers) or characteristics of the 396 client (e.g., the location of the client relative to the servers). 397 Simple redirection is straightforward. The only caveats are (1) 398 there is a limit to the number of alternate IP addresses a single DNS 399 server can manage; and (2) DNS responses are cached by downstream 400 servers so the TTL on the response must be set to an appropriate 401 value so as to preserve the fresheness of the redirection. 403 In CNAME-based DNS redirection, the authoritative server returns a 404 CNAME response to the DNS request, telling the LDNS server to restart 405 the name lookup using a new name. A CNAME is essentially a symbolic 406 link in the DNS namespace, and like a symbolic link, redirection is 407 transparent to the client; the LDNS server gets the CNAME response 408 and re-executes the lookup. Only when the name has been resolved to 409 an IP address does it return the result to the user. Note that DNAME 410 would be preferable to CNAME if it becomes widely supported. 412 2.1.2. HTTP Redirection 414 HTTP redirection makes use of the redirection response of the HTTP 415 protocol (e.g.,"302" or "307"). This response contains a new URL 416 that the application should fetch instead of the original URL. By 417 changing the URL appropriately, the server can cause the user to 418 redirect to a different server. The advantages of HTTP redirection 419 are that (1) the server can change the URL fetched by the client to 420 include, for example, both the DNS name of the particular server to 421 use, as well as the original HTTP server that was being accessed; (2) 422 the client sends the HTTP request to the server, so that its IP 423 address is known and can be used in selecting the server; and (3) 424 other attributes (e.g., content type, user agent type) are visible to 425 the redirection mechanism. 427 The disadvantages of HTTP redirection are (1) it is visible to the 428 application, so it requires application support and may affect the 429 application behavior (e.g., web browsers will not send cookies if the 430 URL changes to a different domain); (2) HTTP is a heavyweight 431 protocol layered on TCP so it has relatively high overhead; and (3) 432 the results of HTTP redirection are not cached so that all 433 redirections must go through to the server. 435 3. Overview of CDNI Operation 437 To provide a big picture overview of the various components of CDN 438 Interconnection, we walk through a "day in the life" of a content 439 item that is made available via a pair of interconnected CDNs. This 440 will serve to illustrate many of the functions that need to be 441 supported in a complete CDNI solution. We give examples using both 442 DNS-based and HTTP-based redirection. We begin with very simple 443 examples and then how additional capabilities, such as recursive 444 request redirection and content removal, might be added. 446 Before walking through some specific examples, we present a high- 447 level view of the operations that may take place. This high-level 448 overview is illustrated in Figure 2. Note that most operations will 449 involve only a subset of all the messages shown below, and that the 450 order and number of operations may vary considerably, as more 451 detailed examples illustrate below. 453 The following shows Operator A as the upstream CDN (uCDN) and 454 Operator B as the downstream CDN (dCDN), where the former has a 455 relationship with a content provider and the latter being the CDN 456 selected by Operator A to deliver content to the end-user. The 457 interconnection relationship may be symmetric between these two CDN 458 operators, but each direction can be considered as operating 459 independently of the other so for simplicity we show the interaction 460 in one direction only. 462 End-User Operator B Operator A 463 | | | 464 | | | 465 | | [Async FCI Push] | (1) 466 | | | 467 | | [MI pre-positioning] | (2) 468 | | | 469 | CONTENT REQUEST | | 470 |-------------------------------------------------->| (3) 471 | | | 472 | | [Sync RI Pull] | (4) 473 | | | 474 | [RI REPLY] | | 475 |<--------------------------------------------------| (5) 476 | | | 477 | | | 478 | CONTENT REQUEST | | 479 |------------------------>| | (6) 480 | | | 481 | | [Sync MI Pull] | (7) 482 | | | 483 | | ACQUISITION REQUEST | 484 | X------------------------>| (8) 485 | X | 486 | X CONTENT DATA | 487 | X<------------------------| (9) 488 | | | 489 | CONTENT DATA | | 490 |<------------------------| | (10) 491 | | | 492 : : : 493 : [Other content requests] : 494 : : : 495 | | [CI: Content Purge] | (11) 496 : : : 497 | | [LI: Log exchange] | (12) 498 | | | 500 Figure 2: Overview of Operation 502 The operations shown in the Figure are as follows: 504 1. dCDN uses the FCI to advertise information relevant to its 505 delivery footprint and capabilities prior to any content 506 requests being redirected. 508 2. Prior to any content request, the uCDN uses the MI to 509 pre=position CDNI metadata to the dCDN, thereby making that 510 metadata available in readiness for later content requests. 512 3. A content request from a user agent arrives at uCDN. 514 4. uCDN may use the RI to synchronously request information from 515 dCDN regarding its delivery capabilities to decide if dCDN is a 516 suitable target for redirection of this request. 518 5. uCDN redirects the request to dCDN by sending some response 519 (DNS, HTTP) to the user agent. 521 6. The user agent requests the content from dCDN. 523 7. dCDN may use the MI to synchronously request metadata related to 524 this content from uCDN, e.g. to decide whether to serve it. 526 8. If the content is not already in a suitable cache in dCDN, dCDN 527 may acquire it from uCDN. 529 9. The content is delivered to dCDN from uCDN. 531 10. The content is delivered to the user agent by dCDN. 533 11. Some time later, perhaps at the request of the CSP (not shown) 534 uCDN may use the CI to instruct dCDN to purge the content, 535 thereby ensuring it is not delivered again. 537 12. After one or more content delivery actions by dCDN, a log of 538 delivery actions may be provided to uCDN using the LI. 540 The following sections show some more specific examples of how these 541 operations may be combined to perform various delivery, control and 542 logging operations across a pair of CDNs. 544 3.1. Preliminaries 546 Initially, we assume that there is at least one CSP that has 547 contracted with an upstream CDN (uCDN) to deliver content on its 548 behalf. We are not particularly concerned with the interface between 549 the CSP and uCDN, other than to note that it is expected to be the 550 same as in the "traditional" (non-interconnected) CDN case. Existing 551 mechanisms such as DNS CNAMEs or HTTP redirects (Section 2) can be 552 used to direct a user request for a piece of content from the CSP 553 towards the CSP's chosen upstream CDN. 555 We assume Operator A provides an upstream CDN that serves content on 556 behalf of a CSP with CDN-Domain cdn.csp.example. We assume that 557 Operator B provides a downstream CDN. An end user at some point 558 makes a request for URL 560 http://cdn.csp.example/...rest of url... 562 It may well be the case that cdn.csp.example is just a CNAME for some 563 other CDN-Domain (such as csp.op-a.example). Nevertheless, the HTTP 564 request in the examples that follow is assumed to be for the example 565 URL above. 567 Our goal is to enable content identified by the above URL to be 568 served by the CDN of operator B. In the following sections we will 569 walk through some scenarios in which content is served, as well as 570 other CDNI operations such as the removal of content from a 571 downstream CDN. 573 3.2. Iterative HTTP Redirect Example 575 In this section we walk through a simple, illustrative example using 576 HTTP redirection from uCDN to dCDN. The example also assumes the use 577 of HTTP redirection inside uCDN and dCDN; however, this is 578 independent of the choice of redirection approach across CDNs, so an 579 alternative example could be constructed still showing HTTP 580 redirection from uCDN to dCDN but using DNS for handling of request 581 inside each CDN. 583 We assume for this example that Operators A and B have established an 584 agreement to interconnect their CDNs, with A being upstream and B 585 being downstream. 587 The operators agree that a CDN-Domain peer-a.op-b.example will be 588 used as the target of redirections from uCDN to dCDN. We assume the 589 name of this domain is communicated by some means to each CDN. (This 590 could be established out-of-band or via a CDNI interface.) We refer 591 to this domain as a "distinguished" CDN-Domain to convey the fact 592 that its use is limited to the interconnection mechanism; such a 593 domain is never used directly by a CSP. 595 We assume the operators also agree on some distinguished CDN-Domain 596 that will be used for inter-CDN acquisition of CSP's content from 597 uCDN by dCDN. In this example, we'll use op-b-acq.op-a.example. 599 We assume the operators also exchange information regarding which 600 requests dCDN is prepared to serve. For example, dCDN may be 601 prepared to serve requests from clients in a given geographical 602 region or a set of IP address prefixes. This information may again 603 be provided out of band or via a defined CDNI interface. 605 We assume DNS is configured in the following way: 607 o The content provider is configured to make operator A the 608 authoritative DNS server for cdn.csp.example (or to return a CNAME 609 for cdn.csp.example for which operator A is the authoritative DNS 610 server). 612 o Operator A is configured so that a DNS request for 613 op-b-acq.op-a.example returns a request router in Operator A. 615 o Operator B is configured so that a DNS request for 616 peer-a.op-b.example/cdn.csp.example returns a request router in 617 Operator B. 619 Figure 3 illustrates how a client request for 621 http://cdn.csp.example/...rest of url... 623 is handled. 625 End-User Operator B Operator A 626 |DNS cdn.csp.example | | 627 |-------------------------------------------------->| 628 | | |(1) 629 |IPaddr of A's Request Router | 630 |<--------------------------------------------------| 631 |HTTP cdn.csp.example | | 632 |-------------------------------------------------->| 633 | | |(2) 634 |302 peer-a.op-b.example/cdn.csp.example | 635 |<--------------------------------------------------| 636 |DNS peer-a.op-b.example | | 637 |------------------------>| | 638 | |(3) | 639 |IPaddr of B's Request Router | 640 |<------------------------| | 641 | | | 642 |HTTP peer-a.op-b.example/cdn.csp.example | 643 |------------------------>| | 644 | |(4) | 645 |302 node1.peer-a.op-b.example/cdn.csp.example | 646 |<------------------------| | 647 |DNS node1.peer-a.op-b.example | 648 |------------------------>| | 649 | |(5) | 650 |IPaddr of B's Delivery Node | 651 |<------------------------| | 652 | | | 653 |HTTP node1.peer-a.op-b.example/cdn.csp.example | 654 |------------------------>| | 655 | |(6) | 656 | |DNS op-b-acq.op-a.example| 657 | |------------------------>| 658 | | |(7) 659 | |IPaddr of A's Request Router 660 | |<------------------------| 661 | |HTTP op-b-acq.op-a.example 662 | |------------------------>| 663 | | |(8) 664 | |302 node2.op-b.acq.op-A.example 665 | |<------------------------| 666 | |DNS node2.op-b-acq.op-a.example 667 | |------------------------>| 668 | | |(9) 669 | |IPaddr of A's Delivery Node 670 | |<------------------------| 671 | | | 672 | |HTTP node2.op-b-acq.op-a.example 673 | |------------------------>| 674 | | |(10) 675 | |Data | 676 | |<------------------------| 677 |Data | | 678 |<------------------------| | 680 Figure 3: Message Flow for Iterative HTTP Redirection 682 The steps illustrated in the figure are as follows: 684 1. A DNS resolver for Operator A processes the DNS request for its 685 customer based on CDN-Domain cdn.csp.example. It returns the IP 686 address of a request router in Operator A. 688 2. A Request Router for Operator A processes the HTTP request and 689 recognizes that the end-user is best served by another CDN, 690 specifically one provided by Operator B, and so it returns a 302 691 redirect message for a new URL constructed by "stacking" 692 Operator B's distinguished CDN-Domain (peer-a.op-b.example) on 693 the front of the original URL. (Note that more complex URL 694 manipulations are possible, such as replacing the initial CDN- 695 Domain by some opaque handle.) 697 3. The end-user does a DNS lookup using Operator B's distinguished 698 CDN-Domain (peer-a.op-b.example). B's DNS resolver returns the 699 IP address of a request router for Operator B. Note that if 700 request routing within dCDN was performed using DNS instead of 701 HTTP redirection, B's DNS resolver would also behave as the 702 request router and directly return the IP address of a delivery 703 node. 705 4. The request router for Operator B processes the HTTP request and 706 selects a suitable delivery node to serve the end-user request, 707 and returns a 302 redirect message for a new URL constructed by 708 replacing the hostname with a subdomain of the Operator B's 709 distinguished CDN-Domain that points to the selected delivery 710 node. 712 5. The end-user does a DNS lookup using Operator B's delivery node 713 subdomain (node1.peer-a.op-b.example). B's DNS resolver returns 714 the IP address of the delivery node. 716 6. The end-user requests the content from B's delivery node. In 717 the case of a cache hit, steps 6, 7, 8, 9 and 10 below do not 718 happen, and the content data is directly returned by the 719 delivery node to the end-user. In the case of a cache miss, the 720 content needs to be acquired by dCDN from uCDN (not the CSP). 721 The distinguished CDN-Domain peer-a.op-b.example indicates to 722 dCDN that this content is to be acquired from uCDN; stripping 723 the CDN-Domain reveals the original CDN-Domain cdn.csp.example 724 and dCDN may verify that this CDN-Domain belongs to a known peer 725 (so as to avoid being tricked into serving as an open proxy). 726 It then does a DNS request for an inter-CDN acquisition CDN- 727 Domain as agreed above (in this case, op-b-acq.op-a.example). 729 7. Operator A's DNS resolver processes the DNS request and returns 730 the IP address of a request router in operator A. 732 8. The request router for Operator A processes the HTTP request 733 from Operator B delivery node. Operator A request router 734 recognizes that the request is from a peer CDN rather than an 735 end-user because of the dedicated inter-CDN acquisition domain 736 (op-b-acq.op-a.example). (Note that without this specially 737 defined inter-CDN acquisition domain, operator A would be at 738 risk of redirecting the request back to operator B, resulting in 739 an infinite loop). The request router for Operator A selects a 740 suitable delivery node in uCDN to serve the inter-CDN 741 acquisition request and returns a 302 redirect message for a new 742 URL constructed by replacing the hostname with a subdomain of 743 the Operator A's distinguished inter-CDN acquisition domain that 744 points to the selected delivery node. 746 9. Operator A DNS resolver processes the DNS request and returns 747 the IP address of the delivery node in operator A. 749 10. Operator B requests (acquires) the content from Operator A. 750 Although not shown, Operator A processes the rest of the URL: it 751 extracts information identifying the origin server, validates 752 that this server has been registered, and determines the content 753 provider that owns the origin server. It may also perform its 754 own content acquisition steps if needed before returning the 755 content to dCDN. 757 The main advantage of this design is that it is simple: each CDN need 758 only know the distinguished CDN-Domain for each peer, with the 759 upstream CDN "pushing" the downstream CDN-Domain onto the URL as part 760 of its redirect (step 2) and the downstream CDN "popping" its CDN- 761 Domain off the URL to expose a CDN-Domain that the upstream CDN can 762 correctly process. Neither CDN needs to be aware of the internal 763 structure of the other's URLs. Moreover, the inter-CDN redirection 764 is entirely supported by a single HTTP redirect; neither CDN needs to 765 be aware of the other's internal redirection mechanism (i.e., whether 766 it is DNS or HTTP based). 768 One disadvantage is that the end-user's browser is redirected to a 769 new URL that is not in the same domain of the original URL. This has 770 implications on a number of security or validation mechanisms 771 sometimes used on endpoints. For example, it is important that any 772 redirected URL be in the same domain (e.g., csp.example) if the 773 browser is expected to send any cookies associated with that domain. 774 As another example, some video players enforce validation of a cross 775 domain policy that needs to allow for the domains involved in the CDN 776 redirection. These problems are generally soluble, but the solutions 777 complicate the example, so we do not discuss them further in this 778 version of the draft. 780 We note that this example begins to illustrate some of the interfaces 781 that may be required for CDNI, but does not require all of them. For 782 example, obtaining information from dCDN regarding the set of client 783 IP addresses or geographic regions it might be able to serve is an 784 aspect of request routing (specifically of the CDNI Footprint & 785 Capabilities Advertisement interface). Important configuration 786 information such as the distinguished names used for redirection and 787 inter-CDN acquisition could also be conveyed via a CDNI interface 788 (e.g., perhaps the CDNI Control interface). The example also shows 789 how existing HTTP-based methods suffice for the acquisition 790 interface. Arguably, the absolute minimum metadata required for CDNI 791 is the information required to acquire the content, and this 792 information was provided "in-band" in this example by means of the 793 URI handed to the client in the HTTP 302 response. The example also 794 assumes that the CSP does not require any distribution policy (e.g. 795 time window, geo-blocking) or delivery processing to be applied by 796 the interconnected CDNs. Hence, there is no explicit CDNI Metadata 797 interface invoked in this example. There is also no explicit CDNI 798 Logging interface discussed in this example. 800 We also note that the step of deciding when a request should be 801 redirected to dCDN rather than served by uCDN has been somewhat 802 glossed over. It may be as simple as checking the client IP address 803 against a list of prefixes, or it may be considerably more complex, 804 involving a wide range of factors, such as the geographic location of 805 the client (perhaps determined from a third party service), CDN load, 806 or specific business rules. 808 This example uses the "iterative" CDNI request redirection approach. 809 That is, uCDN performs part of the request redirection function by 810 redirecting the client to a request router in the dCDN, which then 811 performs the rest of the redirection function by redirecting to a 812 suitable surrogate. If request routing is performed in the dCDN 813 using HTTP redirection, this translates in the end-user experiencing 814 two successive HTTP redirections. By contrast, the alternative 815 approach of "recursive" CDNI request redirection effectively 816 coalesces these two successive HTTP redirections into a single one, 817 sending the end-user directly to the right delivery node in the dCDN. 818 This "recursive" CDNI request routing approach is discussed in the 819 next section. 821 3.3. Recursive HTTP Redirection Example 823 The following example builds on the previous one to illustrate the 824 use of the request routing interface (specifically the CDNI Request 825 Routing Redirection interface) to enable "recursive" CDNI request 826 routing. We build on the HTTP-based redirection approach because it 827 illustrates the principles and benefits clearly, but it is equally 828 possible to perform recursive redirection when DNS-based redirection 829 is employed. 831 In contrast to the prior example, the operators need not agree in 832 advance on a CDN-Domain to serve as the target of redirections from 833 uCDN to dCDN. We assume that the operators agree on some 834 distinguished CDN-Domain that will be used for inter-CDN acquisition 835 of CSP's content by dCDN. In this example, we'll use 836 op-b-acq.op-a.example. 838 We assume the operators also exchange information regarding which 839 requests dCDN is prepared to serve. For example, dCDN may be 840 prepared to serve requests from clients in a given geographical 841 region or a set of IP address prefixes. This information may again 842 be provided out of band or via a defined protocol. 844 We assume DNS is configured in the following way: 846 o The content provider is configured to make operator A the 847 authoritative DNS server for cdn.csp.example (or to return a CNAME 848 for cdn.csp.example for which operator A is the authoritative DNS 849 server). 851 o Operator A is configured so that a DNS request for 852 op-b-acq.op-a.example returns a request router in Operator A. 854 o Operator B is configured so that a request for node1.op-b.example/ 855 cdn.csp.example returns the IP address of a delivery node. Note 856 that there might be a number of such delivery nodes. 858 Figure 3 illustrates how a client request for 860 http://cdn.csp.example/...rest of url... 862 is handled. 864 End-User Operator B Operator A 865 |DNS cdn.csp.example | | 866 |-------------------------------------------------->| 867 | | |(1) 868 |IPaddr of A's Request Router | 869 |<--------------------------------------------------| 870 |HTTP cdn.csp.example | | 871 |-------------------------------------------------->| 872 | | |(2) 873 | |RR/RI REQ cdn.csp.example| 874 | |<------------------------| 875 | | | 876 | |RR/RI RESP node1.op-b.example 877 | |------------------------>| 878 | | |(3) 879 |302 node1.op-b.example/cdn.csp.example | 880 |<--------------------------------------------------| 881 |DNS node1.op-b.example | | 882 |------------------------>| | 883 | |(4) | 884 |IPaddr of B's Delivery Node | 885 |<------------------------| | 886 |HTTP node1.op-b.example/cdn.csp.example | 887 |------------------------>| | 888 | |(5) | 889 | |DNS op-b-acq.op-a.example| 890 | |------------------------>| 891 | | |(6) 892 | |IPaddr of A's Request Router 893 | |<------------------------| 894 | |HTTP op-b-acq.op-a.example 895 | |------------------------>| 896 | | |(7) 897 | |302 node2.op-b.acq.op-A.example 898 | |<------------------------| 899 | |DNS node2.op-b-acq.op-a.example 900 | |------------------------>| 901 | | |(8) 902 | |IPaddr of A's Delivery Node 903 | |<------------------------| 904 | | | 905 | |HTTP node2.op-b-acq.op-a.example 906 | |------------------------>| 907 | | |(9) 908 | |Data | 909 | |<------------------------| 910 |Data | | 911 |<------------------------| | 913 Figure 4: Message Flow for Recursive HTTP Redirection 915 The steps illustrated in the figure are as follows: 917 1. A DNS resolver for Operator A processes the DNS request for its 918 customer based on CDN-Domain cdn.csp.example. It returns the IP 919 address of a Request Router in Operator A. 921 2. A Request Router for Operator A processes the HTTP request and 922 recognizes that the end-user is best served by another CDN-- 923 specifically one provided by Operator B--and so it queries the 924 CDNI Request Routing Redirection interface of Operator B, 925 providing a set of information about the request including the 926 URL requested. Operator B replies with the DNS name of a 927 delivery node. 929 3. Operator A returns a 302 redirect message for a new URL obtained 930 from the RI. 932 4. The end-user does a DNS lookup using the host name of the URL 933 just provided (node1.op-b.example). B's DNS resolver returns the 934 IP address of the corresponding delivery node. Note that, since 935 the name of the delivery node was already obtained from B using 936 the RI, there should not be any further redirection here (in 937 contrast to the iterative method described above.) 939 5. The end-user requests the content from B's delivery node, 940 potentially resulting in a cache miss. In the case of a cache 941 miss, the content needs to be acquired from uCDN (not the CSP.) 942 The distinguished CDN-Domain op-b.example indicates to dCDN that 943 this content is to be acquired from another CDN; stripping the 944 CDN-Domain reveals the original CDN-Domain cdn.csp.example, dCDN 945 may verify that this CDN-Domain belongs to a known peer (so as to 946 avoid being tricked into serving as an open proxy). It then does 947 a DNS request for the inter-CDN Acquisition "distinguished" CDN- 948 Domain as agreed above (in this case, op-b-acq.op-a.example). 950 6. Operator A DNS resolver processes the DNS request and returns the 951 IP address of a request router in operator A. 953 7. The request router for Operator A processes the HTTP request from 954 Operator B delivery node. Operator A request router recognizes 955 that the request is from a peer CDN rather than an end-user 956 because of the dedicated inter-CDN acquisition domain 957 (op-b-acq.op-a.example). (Note that without this specially 958 defined inter-CDN acquisition domain, operator A would be at risk 959 of redirecting the request back to operator B, resulting in an 960 infinite loop). The request router for Operator A selects a 961 suitable delivery node in uCDN to serve the inter-CDN acquisition 962 request and returns a 302 redirect message for a new URL 963 constructed by replacing the hostname with a subdomain of the 964 Operator A's distinguished inter-CDN acquisition domain that 965 points to the selected delivery node. 967 8. Operator A recognizes that the DNS request is from a peer CDN 968 rather than an end-user (due to the internal CDN-Domain) and so 969 returns the address of a delivery node. (Note that without this 970 specially defined internal domain, Operator A would be at risk of 971 redirecting the request back to Operator B, resulting in an 972 infinite loop.) 974 9. Operator B requests (acquires) the content from Operator A. 975 Operator A serves content for the requested CDN-Domain to dCDN. 976 Although not shown, it is at this point that Operator A processes 977 the rest of the URL: it extracts information identifying the 978 origin server, validates that this server has been registered, 979 and determines the content provider that owns the origin server. 980 It may also perform its own content acquisition steps if needed 981 before returning the content to dCDN. 983 Recursive redirection has the advantage over iterative of being more 984 transparent from the end-user's perspective, but the disadvantage of 985 each CDN exposing more of its internal structure (in particular, the 986 addresses of edge caches) to peer CDNs. By contrast, iterative 987 redirection does not require dCDN to expose the addresses of its edge 988 caches to uCDN. 990 This example happens to use HTTP-based redirection in both CDN A and 991 CDN B, but a similar example could be constructed using DNS-based 992 redirection in either CDN. Hence, the key point to take away here is 993 simply that the end user only sees a single redirection of some type, 994 as opposed to the pair of redirections in the prior (iterative) 995 example. 997 The use of the RI requires that the request routing mechanism be 998 appropriately configured and bootstrapped, which is not shown here. 999 More discussion on the bootstrapping of interfaces is provided in 1000 Section 4 1002 3.4. Iterative DNS-based Redirection Example 1004 In this section we walk through a simple example using DNS-based 1005 redirection for request redirection from uCDN to dCDN (as well as for 1006 request routing inside dCDN and uCDN). As noted in Section 2.1, DNS- 1007 based redirection has certain advantages over HTTP-based redirection 1008 (notably, it is transparent to the end-user) as well as some 1009 drawbacks (notably the client IP address is not visible to the 1010 request router). 1012 As before, Operator A has to learn the set of requests that dCDN is 1013 willing or able to serve (e.g. which client IP address prefixes or 1014 geographic regions are part of the dCDN footprint). We assume 1015 Operator has and makes known to operator A some unique identifier 1016 that can be used for the construction of a distinguished CDN-Domain, 1017 as shown in more detail below. (This identifier strictly needs only 1018 to be unique within the scope of Operator A, but a globally unique 1019 identifier, such as an AS number assigned to B, is one easy way to 1020 achieve that.) Also, Operator A obtains the NS records for Operator 1021 B's externally visible redirection servers. Also, as before, a 1022 distinguished CDN-Domain, such as op-b-acq.op-a.example, must be 1023 assigned for inter-CDN acquisition. 1025 We assume DNS is configured in the following way: 1027 o The CSP is configured to make Operator A the authoritative DNS 1028 server for cdn.csp.example (or to return a CNAME for 1029 cdn.csp.example for which operator A is the authoritative DNS 1030 server). 1032 o When uCDN sees a request best served by dCDN, it returns CNAME and 1033 NS records for "b.cdn.csp.example", where "b" is the unique 1034 identifier assigned to Operator B. (It may, for example, be an AS 1035 number assigned to Operator B.) 1037 o dCDN is configured so that a request for "b.cdn.csp.example" 1038 returns a delivery node in dCDN. 1040 o uCDN is configured so that a request for "op-b-acq.op-a.example" 1041 returns a delivery node in uCDN. 1043 Figure 5 depicts the exchange of DNS and HTTP requests. The main 1044 differences from Figure 3 are the lack of HTTP redirection and 1045 transparency to the end-user. 1047 End-User Operator B Operator A 1048 |DNS cdn.csp.example | | 1049 |-------------------------------------------------->| 1050 | | |(1) 1051 |CNAME b.cdn.csp.example | | 1052 |NS records for b.cdn.csp.example | 1053 |<--------------------------------------------------| 1054 |DNS b.cdn.csp.example | | 1055 |------------------------>| | 1056 | |(2) | 1057 |IPaddr of B's Delivery Node | 1058 |<------------------------| | 1059 |HTTP cdn.csp.example | | 1060 |------------------------>| | 1061 | |(3) | 1062 | |DNS op-b-acq.op-a.example| 1063 | |------------------------>| 1064 | | |(4) 1065 | |IPaddr of A's Delivery Node 1066 | |<------------------------| 1067 | |HTTP op-b-acq.op-a.example 1068 | |------------------------>| 1069 | | |(5) 1070 | |Data | 1071 | |<------------------------| 1072 |Data | | 1073 |<------------------------| | 1075 Figure 5: Message Flow for DNS-based Redirection 1077 The steps illustrated in the figure are as follows: 1079 1. Request Router for Operator A processes the DNS request for CDN- 1080 Domain cdn.csp.example and recognizes that the end-user is best 1081 served by another CDN. (This may depend on the IP address of the 1082 user's local DNS resolver, or other information discussed below.) 1083 The Request Router returns a DNS CNAME response by "stacking" the 1084 distinguished identifier for Operator B onto the original CDN- 1085 Domain (e.g., b.cdn.csp.example), plus an NS record that maps 1086 b.cdn.csp.example to B's Request Router. 1088 2. The end-user does a DNS lookup using the modified CDN-Domain 1089 (i.e., b.cdn.csp.example). This causes B's Request Router to 1090 respond with a suitable delivery node. 1092 3. The end-user requests the content from B's delivery node. The 1093 requested URL contains the name cdn.csp.example. (Note that the 1094 returned CNAME does not affect the URL.) At this point the 1095 delivery node has the correct IP address of the end-user and can 1096 do an HTTP 302 redirect if the redirections in steps 2 and 3 were 1097 incorrect. Otherwise B verifies that this CDN-Domain belongs to 1098 a known peer (so as to avoid being tricked into serving as an 1099 open proxy). It then does a DNS request for an "internal" CDN- 1100 Domain as agreed above (op-b-acq.op-a.example). 1102 4. Operator A recognizes that the DNS request is from a peer CDN 1103 rather than an end-user (due to the internal CDN-Domain) and so 1104 returns the address of a delivery node in uCDN. 1106 5. Operator A serves content to dCDN. Although not shown, it is at 1107 this point that Operator A processes the rest of the URL: it 1108 extracts information identifying the origin server, validates 1109 that this server has been registered, and determines the content 1110 provider that owns the origin server. 1112 The advantages of this approach are that it is more transparent to 1113 the end-user and requires fewer round trips than HTTP-based 1114 redirection (in its worst case, i.e., when none of the needed DNS 1115 information is cached). A potential problem is that the upstream CDN 1116 depends on being able to learn the correct downstream CDN that serves 1117 the end-user from the client address in the DNS request. In standard 1118 DNS operation, uCDN will only obtain the address of the client's 1119 local DNS resolver (LDNS), which is not guaranteed to be in the same 1120 network (or geographic region) as the client. If not--e.g., the end- 1121 user uses a global DNS service--then the upstream CDN cannot 1122 determine the appropriate downstream CDN to serve the end-user. In 1123 this case, and assuming the uCDN is capable of detecting that 1124 situation, one option is for the upstream CDN to treat the end-user 1125 as it would any user not connected to a peer CDN. Another option is 1126 for the upstream CDN to "fall back" to a pure HTTP-based redirection 1127 strategy in this case (i.e., use the first method). Note that this 1128 problem affects existing CDNs that rely on DNS to determine where to 1129 redirect client requests, but the consequences are arguably less 1130 serious for CDNI since the LDNS is likely in the same network as the 1131 dCDN serves. One approach to ensuring that the client's IP address 1132 prefix is correctly determined in such situations is described in 1133 [I-D.vandergaast-edns-client-subnet]. 1135 As with the prior example, this example partially illustrates the 1136 various interfaces involved in CDNI. Operator A could learn 1137 dynamically from Operator B the set of prefixes or regions that B is 1138 willing and able to serve via the CDNI Footprint & Capabilities 1139 Advertisement interface. The distinguished name used for acquisition 1140 and the identifier for Operator B that is prepended to the CDN-Domain 1141 on redirection are examples of information elements that might also 1142 be conveyed by CDNI interfaces (or, alternatively, statically 1143 configured). As before, minimal metadata sufficient to obtain the 1144 content is carried "in-band" as part of the redirection process, and 1145 standard HTTP is used for inter-CDN acquisition. There is no 1146 explicit CDNI Logging interface discussed in this example. 1148 3.5. Dynamic Footprint Discovery Example 1150 There could be situations where being able to dynamically discover 1151 the set of requests that a given dCDN is willing and able to serve is 1152 beneficial. For example, a CDN might at one time be able to serve a 1153 certain set of client IP prefixes, but that set might change over 1154 time due to changes in the topology and routing policies of the IP 1155 network. The following example illustrates this capability. We have 1156 chosen the example of DNS-based redirection, but HTTP-based 1157 redirection could equally well use this approach. 1159 End-User Operator B Operator A 1160 |DNS cdn.csp.example | | 1161 |-------------------------------------------------->| 1162 | | |(1) 1163 | | RI REQ op-b.example | 1164 | |<------------------------| 1165 | | |(2) 1166 | | RI REPLY | 1167 | |------------------------>| 1168 | | |(3) 1169 |CNAME b.cdn.csp.example | | 1170 |NS records for b.cdn.csp.example | 1171 |<--------------------------------------------------| 1172 |DNS b.cdn.csp.example | | 1173 |------------------------>| | 1174 | |(2) | 1175 |IPaddr of B's Delivery Node | 1176 |<------------------------| | 1177 |HTTP cdn.csp.example | | 1178 |------------------------>| | 1179 | |(3) | 1180 | |DNS op-b-acq.op-a.example| 1181 | |------------------------>| 1182 | | |(4) 1183 | |IPaddr of A's Delivery Node 1184 | |<------------------------| 1185 | |HTTP op-b-acq.op-a.example 1186 | |------------------------>| 1187 | | |(5) 1188 | |Data | 1189 | |<------------------------| 1190 |Data | | 1191 |<------------------------| | 1193 Figure 6: Message Flow for Dynamic Footprint Discovery 1195 This example differs from the one in Figure 5 only in the addition of 1196 a FCI request (step 2) and corresponding response (step 3). The RI 1197 REQ could be a message such as "Can you serve clients from this IP 1198 Prefix?" or it could be "Provide the list of client IP prefixes you 1199 can currently serve". In either case the response might be cached by 1200 operator A to avoid repeatedly asking the same question. 1201 Alternatively, or in addition, Operator B may spontaneously advertise 1202 to Operator A information (or changes) on the set of requests it is 1203 willing and able to serve on behalf of operator A; in that case, 1204 Operator B may spontaneously issue RR/RI REPLY messages that are not 1205 in direct response to a corresponding RR/RI REQ message. (Note that 1206 the issues of determining the client's subnet from DNS requests, as 1207 described above, are exactly the same here as in Section 3.4.) 1209 Once Operator A obtains the RI response, it is now able to determine 1210 that Operator B's CDN is an appropriate dCDN for this request and 1211 therefore a valid candidate dCDN to consider in its Redirection 1212 decision. If that dCDN is selected, the redirection and serving of 1213 the request proceeds as before (i.e. in the absence of dynamic 1214 footprint discovery). 1216 3.6. Content Removal Example 1218 The following example illustrates how the CDNI Control interface may 1219 be used to achieve pre-positioning of an item of content in the dCDN. 1220 In this example, user requests for a particular content, and 1221 corresponding redirection of such requests from Operator A to 1222 Operator B CDN, may (or may not) have taken place earlier. Then, at 1223 some point in time, the uCDN (for example, in response to a 1224 corresponding Trigger from the Content Provider) uses the CI to 1225 request that content identified by a particular URL be removed from 1226 dCDN. The following diagram illustrates the operation. 1228 End-User Operator B Operator A 1229 | |CI purge cdn.csp.example/... 1230 | |<------------------------| 1231 | | | 1232 | |CI OK | 1233 | |------------------------>| 1234 | | | 1236 Figure 7: Message Flow for Content Removal 1238 The CI is used to convey the request from uCDN to dCDN that some 1239 previously acquired content should be deleted. The URL in the 1240 request specifies which content to remove. This example corresponds 1241 to a DNS-based redirection scenario such as Section 3.4. If HTTP- 1242 based redirection had been used, the URL for removal would be of the 1243 form peer-a.op-b.example/cdn.csp.example/... 1245 The dCDN is expected to confirm to the uCDN, as illustrated by the CI 1246 OK message, the completion of the removal of the targeted content 1247 from all the caches in dCDN. 1249 3.7. Pre-Positioned Content Acquisition Example 1251 The following example illustrates how the CI may be used to pre- 1252 position an item of content in the dCDN. In this example, Operator A 1253 uses the CDNI Metadata interface to request that content identified 1254 by a particular URL be pre-positioned into Operator B CDN. 1256 End-User Operator B Operator A 1258 | |CI pre-position cdn.csp.example/... 1259 | |<------------------------| 1260 | | |(1) 1261 | |CI OK | 1262 | |------------------------>| 1263 | | | 1264 | |DNS op-b-acq.op-a.example| 1265 | |------------------------>| 1266 | | |(2) 1267 | |IPaddr of A's Delivery Node 1268 | |<------------------------| 1269 | |HTTP op-b-acq.op-a.example 1270 | |------------------------>| 1271 | | |(3) 1272 | |Data | 1273 | |<------------------------| 1274 |DNS cdn.csp.example | | 1275 |--------------------------------------------->| 1276 | | |(4) 1277 |IPaddr of A's Request Router | 1278 |<---------------------------------------------| 1279 |HTTP cdn.csp.example| | 1280 |--------------------------------------------->| 1281 | | |(5) 1282 |302 peer-a.op-b.example/cdn.csp.example | 1283 |<---------------------------------------------| 1284 |DNS peer-a.op-b.example | 1285 |------------------->| | 1286 | |(6) | 1287 |IPaddr of B's Delivery Node | 1288 |<-------------------| | 1289 |HTTP peer-a.op-b.example/cdn.csp.example | 1290 |------------------->| | 1291 | |(7) | 1292 |Data | | 1293 |<-------------------| | 1295 Figure 8: Message Flow for Content Pre-Positioning 1297 The steps illustrated in the figure are as follows: 1299 1. Operator A uses the CI to request that Operator B pre-positions a 1300 particular content item identified by its URL. Operator B 1301 responds by confirming that it is willing to perform this 1302 operation. 1304 Steps 2 and 3 are exactly the same as steps 5 and 6 of Figure 3, only 1305 this time those steps happen as the result of the Pre-positioning 1306 request instead of as the result of a cache miss. 1308 Steps 4, 5, 6, 7 are exactly the same as steps 1, 2, 3, 4 of 1309 Figure 3, only this time Operator B CDN can serve the end-user 1310 request without triggering dynamic content acquisition, since the 1311 content has been pre-positioned in dCDN. Note that, depending on 1312 dCDN operations and policies, the content pre-positioned in the dCDN 1313 may be pre-positioned to all, or a subset of, dCDN caches. In the 1314 latter case, intra-CDN dynamic content acquisition may take place 1315 inside the dCDN serving requests from caches on which the content has 1316 not been pre-positioning; however, such intra-CDN dynamic acquisition 1317 would not involve the uCDN. 1319 3.8. Asynchronous CDNI Metadata Example 1321 In this section we walk through a simple example illustrating a 1322 scenario of asynchronously exchanging CDNI metadata, where the 1323 downstream CDN obtains CDNI metadata for content ahead of a 1324 corresponding content request. The example that follows assumes that 1325 HTTP-based inter-CDN redirection and recursive CDNI request-routing 1326 are used, as in Section 3.3. However, Asynchronous exchange of CDNI 1327 Metadata is similarly applicable to DNS-based inter-CDN redirection 1328 and iterative request routing (in which cases the CDNI metadata may 1329 be used at slightly different processing stages of the message 1330 flows). 1332 End-User Operator B Operator A 1333 | | | 1334 | |CI pre-position (Trigger)| 1335 | |<------------------------|(1) 1336 | | | 1337 | |CI OK | 1338 | |------------------------>|(2) 1339 | | | 1340 | |MI pull REQ | 1341 | |------------------------>|(3) 1342 | | | 1343 | |MI metadata REP |(4) 1344 | | | 1345 | | | 1346 | CONTENT REQUEST | | 1347 |-------------------------------------------------->|(5) 1348 | | | 1349 | | RI REQ | 1350 | |<------------------------|(6) 1351 | | | 1352 | | RI RESP | 1353 | |------------------------>|(7) 1354 | | | 1355 | CONTENT REDIRECTION | | 1356 |<--------------------------------------------------|(8) 1357 | | | 1358 | CONTENT REQUEST | | 1359 |------------------------>|(9) | 1360 | | | 1361 : : : 1362 | CONTENT DATA | | 1363 |<------------------------| |(10) 1365 Figure 9: Message Flow for Asynchronous CDNI Metadata 1367 The steps illustrated in the figure are as follows: 1369 1. Operator A uses the CI to Trigger to signal the availability of 1370 CDNI metadata to Operator B. 1372 2. Operator B acknowledges the receipt of this Trigger. 1374 3. Operator B requests the latest metadata from Operator A using 1375 the MI. 1377 4. Operator A replies with the requested metadata. This document 1378 does not constrain how the CDNI metadata information is actually 1379 represented. For the purposes of this example, we assume that 1380 Operator A provides CDNI metadata to Operator B indicating that: 1382 * this CDNI Metadata is applicable to any content referenced by 1383 some CDN-Domain. 1385 * this CDNI metadata consists of a distribution policy 1386 requiring enforcement by the delivery node of a specific per- 1387 request authorization mechanism (e.g. URI signature or token 1388 validation). 1390 5. A Content Request occurs as usual. 1392 6. A CDNI Request Routing Redirection request (RI REQ) is issued by 1393 operator A CDN, as discussed in Section 3.3. Operator B's 1394 request router can access the CDNI Metadata that are relevant to 1395 the requested content and that have been pre-positioned as per 1396 Steps 1-4, which may or may not affect the response. 1398 7. Operator B's request router issues a CDNI Request Routing 1399 Redirection response (RI RESP) as in Section 3.3. 1401 8. Operator B performs content redirection as discussed in 1402 Section 3.3. 1404 9. On receipt of the Content Request by the end user, the delivery 1405 node detects that previously acquired CDNI metadata is 1406 applicable to the requested content. In accordance with the 1407 specific CDNI metadata of this example, the delivery node will 1408 invoke the appropriate per-request authorization mechanism, 1409 before serving the content. (Details of this authorization are 1410 not shown.) 1412 10. Assuming successful per-request authorization, serving of 1413 Content Data (possibly preceded by inter-CDN acquisition) 1414 proceeds as in Section 3.3. 1416 3.9. Synchronous CDNI Metadata Acquisition Example 1418 In this section we walk through a simple example illustrating a 1419 scenario of Synchronous CDNI metadata acquisition, in which the 1420 downstream CDN obtains CDNI metadata for content at the time of 1421 handling a first request for the corresponding content. As in the 1422 preceding section, this example assumes that HTTP-based inter-CDN 1423 redirection and recursive CDNI request-routing are used (as in 1424 Section 3.3), but dynamic CDNI metadata acquisition is applicable to 1425 other variations of request routing. 1427 End-User Operator B Operator A 1428 | | | 1429 | CONTENT REQUEST | | 1430 |-------------------------------------------------->|(1) 1431 | | | 1432 | | RI REQ | 1433 | (2)|<------------------------| 1434 | | | 1435 | | MI REQ | 1436 | (3)|------------------------>| 1437 | | MI RESP | 1438 | |<------------------------|(4) 1439 | | | 1440 | | RI RESP | 1441 | |------------------------>|(5) 1442 | | | 1443 | | | 1444 | CONTENT REDIRECTION | | 1445 |<--------------------------------------------------|(6) 1446 | | | 1447 | CONTENT REQUEST | | 1448 |------------------------>|(7) | 1449 | | | 1450 | | MI REQ | 1451 | (8)|------------------------>| 1452 | | MI RESP | 1453 | |<------------------------|(9) 1454 | | | 1455 : : : 1456 | CONTENT DATA | | 1457 |<------------------------| |(10) 1459 Figure 10: Message Flow for Synchronous CDNI Metadata Acquisition 1461 The steps illustrated in the figure are as follows: 1463 1. A Content Request arrives as normal. 1465 2. An RI request occurs as in the prior example. 1467 3. On receipt of the CDNI Request Routing Request, Operator B's CDN 1468 initiates Synchronous acquisition of CDNI Metadata that are 1469 needed for routing of the end-user request. We assume the URI 1470 for the a Metadata server is known ahead of time through some 1471 out-of-band means. 1473 4. On receipt of a CDNI Metadata Request, Operator A's CDN 1474 responds, making the corresponding CDNI metadata information 1475 available to Operator B's CDN. This metadata is considered by 1476 operator B's CDN before responding to the Request Routing 1477 request. (In a simple case, the metadata could simply be an 1478 allow or deny response for this particular request.) 1480 5. Response to the RI request as normal. 1482 6. Redirection message is sent to the end user. 1484 7. A delivery node of Operator B receives the end user request. 1486 8. The delivery node Triggers dynamic acquisition of additional 1487 CDNI metadata that are needed to process the end-user content 1488 request. Note that there may exist cases where this step need 1489 not happen, for example because the metadata were already 1490 acquired previously. 1492 9. Operator A's CDN responds to the CDNI Metadata Request and makes 1493 the corresponding CDNI metadata available to Operator B. This 1494 metadata influence how Operator B's CDN processes the end-user 1495 request. 1497 10. Content is served (possibly preceded by inter-CDN acquisition) 1498 as in Section 3.3. 1500 3.10. Content and Metadata Acquisition with Multiple Upstream CDNs 1502 A single dCDN may receive end-user requests from multiple uCDNs. 1503 When a dCDN receives an end-user request, it must determine the 1504 identity of the uCDN from which it should acquire the requested 1505 content. 1507 Ideally, the acquisition path of an end-user request will follow the 1508 redirection path of the request. The dCDN should acquire the content 1509 from the same uCDN which redirected the request. 1511 Determining the acquisition path requires the dCDN to reconstruct the 1512 redirection path based on information in the end-user request. The 1513 method for reconstructing the redirection path differs based on the 1514 redirection approach: HTTP or DNS. 1516 With HTTP-redirection, the rewritten URI should include sufficient 1517 information for the dCDN to directly or indirectly determine the uCDN 1518 when the end-user request is received. The HTTP-redirection approach 1519 can be further broken-down based on the how the URL is rewritten 1520 during redirection: HTTP-redirection with or without Site 1521 Aggregation. HTTP-redirection with Site Aggregation hides the 1522 identity of the original CSP. HTTP-redirection without Site 1523 Aggregation does not attempt to hide the identity of the original 1524 CSP. With both approaches, the rewritten URI includes enough 1525 information to identify the immediate neighbor uCDN. 1527 With DNS-redirection, the dCDN receives the published URI (instead of 1528 a rewritten URI) and does not have sufficient information for the 1529 dCDN to identify the appropriate uCDN. The dCDN may narrow the set 1530 of viable uCDNs by examining the CDNI metadata from each to determine 1531 which uCDNs are hosting metadata for the requested content. If there 1532 is a single uCDN hosting metadata for the requested content, the dCDN 1533 can assume that the request redirection is coming from this uCDN and 1534 can acquire content from that uCDN. If there are multiple uCDNs 1535 hosting metadata for the requested content, the dCDN may be ready to 1536 trust any of these uCDNs to acquire the content (provided the uCDN is 1537 in a position to serve it). If the dCDN is not ready to trust any of 1538 these uCDNs, it needs to ensure via out of band arrangements that, 1539 for a given content, only a single uCDN will ever redirect requests 1540 to the dCDN. 1542 Content acquisition may be preceded by content metadata acquisition. 1543 If possible, the acquisition path for metadata should also follow the 1544 redirection path. Additionally, we assume metadata is indexed based 1545 on rewritten URIs in the case of HTTP-redirection and is indexed 1546 based on published URIs in the case of DNS-redirection. Thus, the RI 1547 and the MI are tightly coupled in that the result of request routing 1548 (a rewritten URI pointing to the dCDN) serves as an input to metadata 1549 lookup. If the content metadata includes information for acquiring 1550 the content, then the MI is also tightly coupled with the acquisition 1551 interface in that the result of the metadata lookup (an acquisition 1552 URL likely hosted by the uCDN) should serve as input to the content 1553 acquisition. 1555 4. Main Interfaces 1557 Figure 1 illustrates the main interfaces that are in scope for the 1558 CDNI WG, along with several others. The detailed specifications of 1559 these interfaces are left to other documents, but see [RFC6707] and 1560 [I-D.ietf-cdni-requirements] for some discussion of the interfaces. 1562 One interface that is not shown in Figure 1 is the interface between 1563 the user and the CSP. While for the purposes of CDNI that interface 1564 is out of scope, it is worth noting that it does exist and can 1565 provide useful functions, such as end-to-end performance monitoring 1566 and some forms of authentication and authorization. 1568 There is also an important interface between the user and the Request 1569 Routing function of both uCDN and dCDN (shown as the "Request" 1570 Interface in Figure 1). As we saw in some of the preceding examples, 1571 that interface can be used as a way of passing information a subset 1572 of metadata such as the minimum information that is required for dCDN 1573 to obtain the content from uCDN. 1575 In this section we will provide an overview of the functions 1576 performed by each of the CDNI interfaces and discuss how they fit 1577 into the overall solution. We also examine some of the design 1578 tradeoffs, and explore several cross-interface concerns. We begin 1579 with an examination of one such tradeoff that affects all the 1580 interfaces - the use of in-band or out-of-band communication. 1582 4.1. In-Band versus Out-of-Band Interfaces 1584 Before getting to the individual interfaces, we observe that there is 1585 a high-level design choice for each, involving the use of existing 1586 in-band communication channels versus defining new out-of-band 1587 interfaces. 1589 It is possible that the information needed to carry out various 1590 interconnection functions can be communicated between peer CDNs using 1591 existing in-band protocols. The use of HTTP 302 redirect is an 1592 example of how certain aspects of request routing can be implemented 1593 in-band (embedded in URIs). Note that using existing in-band 1594 protocols does not imply that the CDNI interfaces are null; it is 1595 still necessary to establish the rules (conventions) by which such 1596 protocols are used to implement the various interface functions. 1598 There are other opportunities for in-band communication beyond HTTP 1599 redirects. For example, many of the HTTP directives used by proxy 1600 servers can also be used by peer CDNs to inform each other of caching 1601 activity. Of these, one that is particularly relevant is the If- 1602 Modified-Since directive, which is used with the GET method to make 1603 it conditional: if the requested object has not been modified since 1604 the time specified in this field, a copy of the object will not be 1605 returned, and instead, a 304 (not modified) response will be 1606 returned. 1608 4.2. Cross Interface Concerns 1610 Although the CDNI interfaces are largely independent, there are a set 1611 of conventions practiced consistently across all interfaces. Most 1612 important among these is how resources are named, for exampmle, how 1613 the CDNI Metadata and Control interfaces identify the set of 1614 resources to which a given directive applies, or the CDNI Logging 1615 interface identifies the set of resources for which a summary record 1616 applies. 1618 While in the limit the CDNI interfaces could explicitly identify 1619 every individual resource, in practice, they name resource aggregates 1620 (sets of URIs) that are to be treated in a similar way. For example, 1621 URI aggregates can be identified by a CDN-Domain (i.e., the FQDN at 1622 the beginning of a URI) or by a URI-Filter (i.e., a regular 1623 expression that matches a subset of URIs contained in some CDN- 1624 Doman). In other words, CDN-Domains and URI-Filters provide a 1625 uniform means to aggregate sets (and subsets) of URIs for the purpose 1626 of defining the scope for some operation in one of the CDNI 1627 interfaces. 1629 4.3. Request Routing Interfaces 1631 The Request Routing interface comprises two parts: the Asynchronous 1632 interface used by a dCDN to advertize footprint and capabilities 1633 (denoted FCI) to a uCDN, allowing the uCDN to decide whether to 1634 redirect particular user requests to that dCDN; and the Synchronous 1635 interface used by the uCDN to redirect a user request to the dCDN 1636 (denoted RI). (These are somewhat analogous to the operations of 1637 routing and forwarding in IP.) 1639 As illustrated in Section 3, the RI part of request routing may be 1640 implemented in part by DNS and HTTP. Naming conventions may be 1641 established by which CDN peers communicate whether a request should 1642 be routed or content served. 1644 We also note that RI plays a key role in enabling recursive 1645 redirection, as illustrated in Section 3.3. It enables the user to 1646 be redirected to the correct delivery node in dCDN with only a single 1647 redirection step (as seen by the user). This may be particularly 1648 valuable as the chain of interconnected CDNs increases beyond two 1649 CDNs. For further discussion on the RI, see 1650 [I-D.ietf-cdni-redirection]. 1652 In support of these redirection requests, it is necessary for CDN 1653 peers to exchange additional information with each other, and this is 1654 the role of the FCI part of request routing. Depending on the 1655 method(s) supported, this might includes 1657 o The operator's unique id (operator-id) or distinguished CDN-Domain 1658 (operator-domain); 1660 o NS records for the operator's set of externally visible request 1661 routers; 1663 o The set of requests the dCDN operator is prepared to serve (e.g. a 1664 set of client IP prefixes or geographic regions that may be served 1665 by dCDN). 1667 o Additional capabilities of the dCDN, such as its ability to 1668 support different CDNI Metadata requests. 1670 Note that the set of requests that dCDN is willing to serve could in 1671 some cases be relatively static (e.g., a set of IP prefixes) which 1672 could be exchanged off-line, or might even be negotiated as part of a 1673 peering agreement. However, it may also be more dynamic, in which 1674 case the exchange supported by FCI would be be helpful. A further 1675 discussion of the Footprint & Capability Advertisement interface can 1676 be found in [I-D.ietf-cdni-footprint-capabilities-semantics]. 1678 4.4. CDNI Logging Interface 1680 It is necessary for the upstream CDN to have visibility into the 1681 delivery of content that it redirected to a downstream CDN. This 1682 allows the upstream CDN to properly bill its customers for multiple 1683 deliveries of content cached by the downstream CDN, as well as to 1684 report accurate traffic statistics to those content providers. This 1685 is one role of the LI. 1687 Other operational data that may be relevant to CDNI can also be 1688 exchanged by the LI. For example, dCDN may report the amount of 1689 content it has acquired from uCDN, and how much cache storage has 1690 been consumed by content cached on behalf of uCDN. 1692 Traffic logs are easily exchanged off-line. For example, the 1693 following traffic log is a small deviation from the Apache log file 1694 format, where entries include the following fields: 1696 o Domain - the full domain name of the origin server 1698 o IP address - the IP address of the client making the request 1700 o End time - the ending time of the transfer 1702 o Time zone - any time zone modifier for the end time 1704 o Method - the transfer command itself (e.g., GET, POST, HEAD) 1706 o URL - the requested URL 1708 o Version - the protocol version, such as HTTP/1.0 1710 o Response - a numeric response code indicating transfer result 1711 o Bytes Sent - the number of bytes in the body sent to the client 1713 o Request ID - a unique identifier for this transfer 1715 o User agent - the user agent, if supplied 1717 o Duration - the duration of the transfer in milliseconds 1719 o Cached Bytes - the number of body bytes served from the cache 1721 o Referrer - the referrer string from the client, if supplied 1723 Of these, only the Domain field is indirect in the downstream CDN--it 1724 is set to the CDN-Domain used by the upstream CDN rather than the 1725 actual origin server. This field could then used to filter traffic 1726 log entries so only those entries matching the upstream CDN are 1727 reported to the corresponding operator. Further discussion of the LI 1728 can be found in [I-D.ietf-cdni-logging]. 1730 One open question is who does the filtering. One option is that the 1731 downstream CDN filters its own logs, and passes the relevant records 1732 directly to each upstream peer. This requires that the downstream 1733 CDN knows the set of CDN-Domains that belong to each upstream peer. 1734 If this information is already exchanged between peers as part of 1735 another interface, then direct peer-to-peer reporting is 1736 straightforward. If it is not available, and operators do not wish 1737 to advertise the set of CDN-Domains they serve to their peers, then 1738 the second option is for each CDN to send both its non-local traffic 1739 records and the set of CDN-Domains it serves to an independent third- 1740 party (i.e., a CDN Exchange), which subsequently filters, merges, and 1741 distributes traffic records on behalf of each participating CDN 1742 operator. 1744 A second open question is how timely traffic information should be. 1745 For example, in addition to offline traffic logs, accurate real-time 1746 traffic monitoring might also be useful, but such information 1747 requires that the downstream CDN inform the upstream CDN each time it 1748 serves upstream content from its cache. The downstream CDN can do 1749 this, for example, by sending a conditional HTTP GET request (If- 1750 Modified-Since) to the upstream CDN each time it receives an HTTP GET 1751 request from one of its end-users. This allows the upstream CDN to 1752 record that a request has been issued for the purpose of real-time 1753 traffic monitoring. The upstream CDN can also use this information 1754 to validate the traffic logs received later from the downstream CDN. 1756 There is obviously a tradeoff between accuracy of such monitoring and 1757 the overhead of the downstream CDN having to go back to the upstream 1758 CDN for every request. 1760 Another design tradeoff in the LI is the degree of aggregation or 1761 summarization of data. One situation that lends itself to 1762 summarization is the delivery of HTTP adaptive streaming (HAS), since 1763 the large number of individual chunk requests potentially results in 1764 large volumes of logging information. This case is discussed below, 1765 but other forms of aggregation may also be useful. For example, 1766 there may be situations where bulk metrics such as bytes delivered 1767 per hour may suffice rather than the detailed per-request logs 1768 outlined above. It seems likely that a range of granularities of 1769 logging will be needed along with ways to specify the type and degree 1770 of aggregation required. 1772 4.5. CDNI Control Interface 1774 The CDNI Control interface is initially used to bootstrap the other 1775 interfaces. As a simple example, it could be used to provide the 1776 address of the logging server in dCDN to uCDN in order to bootstrap 1777 the CDNI Logging interface. It may also be used, for example, to 1778 establish security associations for the other interfaces. 1780 The other role the CI plays is to allow the uCDN to pre-position, 1781 revalidate, or purge metadata and content on a dCDN. These 1782 operations, sometimes collectively called the Trigger interface, are 1783 discussed further in [I-D.ietf-cdni-control-triggers]. 1785 4.6. CDNI Metadata Interface 1787 The role of the CDNI Metadata interface is to enable CDNI 1788 distribution metadata to be conveyed to the downstream CDN by the 1789 upstream CDN. Such metadata includes geo-blocking restrictions, 1790 availability windows, access control policies, and so on. It may 1791 also include information to facilitate acquisition of content by dCDN 1792 (e.g., alternate sources for the content, authorization information 1793 needed to acquire the content from the source). For a full 1794 discussion of the CDNI Metadata Interface, see 1795 [I-D.ietf-cdni-metadata] 1797 Some distribution metadata may be partially emulated using in-band 1798 mechanisms. For example, in case of any geo-blocking restrictions or 1799 availability windows, the upstream CDN can elect to redirect a 1800 request to the downstream CDN only if that CDN's advertised delivery 1801 footprint is acceptable for the requested URL. Similarly, the 1802 request could be forwarded only if the current time is within the 1803 availability window. However, such approaches typically come with 1804 shortcomings such as inability to prevent from replay outside the 1805 time window or inability to make use of a downstream CDN that covers 1806 a broader footprint than the geo-blocking restrictions. 1808 Similarly, some forms of access control may also be performed on a 1809 per-request basis using HTTP directives. For example, being able to 1810 respond to a conditional GET request gives the upstream CDN an 1811 opportunity to influence how the downstream CDN delivers its content. 1812 Minimally, the upstream CDN can invalidate (purge) content previously 1813 cached by the downstream CDN. 1815 Fine-grain control over how the downstream CDN delivers content on 1816 behalf of the upstream CDN is also possible. For example, by 1817 including the Forwarded HTTP header [I-D.ietf-appsawg-http-forwarded] 1818 with the conditional GET request, the downstream CDN can report the 1819 end-user's IP address to the upstream CDN, giving it an opportunity 1820 to control whether the downstream CDN should serve the content to 1821 this particular end-user. The upstream CDN would communicate its 1822 directive through its response to the conditional GET. The 1823 downstream CDN can cache information for a period of time specified 1824 by the upstream CDN, thereby reducing control overhead, but then 1825 preventing per-request control during the corresponding caching 1826 period. 1828 All of these in-band techniques serve to illustrate that uCDNs have 1829 the option of enforcing some of their access control policies 1830 themselves (at the expense of increased inter-CDN signaling load), 1831 rather than delegating enforcement to dCDNs using the MI. As a 1832 consequence, the MI could provide a means for the uCDN to express its 1833 desire to retain enforcement for itself. For example, this might be 1834 done by including a "check with me" flag in the metadata associated 1835 with certain content. The realization of such in-band techniques 1836 over the various inter-CDN acquisition protocols (e.g., HTTP) 1837 requires further investigation and may require small extensions or 1838 semantic changes to the acquisition protocol. 1840 4.7. HTTP Adaptive Streaming Concerns 1842 We consider HTTP Adaptive Streaming (HAS) and the impact it has on 1843 the CDNI interfaces because large objects (e.g., videos) are broken 1844 into a sequence of small, independent chunks. For each of the 1845 following, a more thorough discussion, including an overview of the 1846 tradeoffs involved in alternative designs, can be found in RFC 6983. 1848 First, with respect to Content Acquisition and File Management, which 1849 are out-of-scope for the CDNI interfaces but nontheless relevant to 1850 the overall operation, we assume no additional measures are required 1851 to deal with large numbers of chunks. This means that the dCDN is 1852 not explicitly made aware of any relationship between different 1853 chunks and the dCDN handles each chunk as if it were an individual 1854 and independent content item. The result is that content acquisition 1855 between uCDN and dCDN also happens on a per-chunk basis. This 1856 approach is in line with the recommendations made in RFC 6983, which 1857 also identifies potential improvements in this area that might be 1858 considered in the future. 1860 Second, with respect to Request Routing, we note that HAS manifest 1861 files have the potential to interfere with request routing since 1862 manifest files contain URLs pointing to the location of content 1863 chunks. To make sure that a manifest file does not hinder CDNI 1864 request routing and does not place excessive load on CDNI resources, 1865 the use of manifest files could either be limited to those containing 1866 relative URLs or the uCDN could modify the URLs in the manifest. Our 1867 approach for dealing with these issues is twofold. As a mandatory 1868 requirement, CDNs should be able to handle unmodified manifest files 1869 containing either relative or absolute URLs. To limit the number of 1870 redirects, and thus the load placed on the CDNI interfaces, as an 1871 optional feature uCDNs can use the information obtained through the 1872 CNDI Request Routing Redirection interface to modify the URLs in the 1873 manifest file. Since the modification of the manifest file is an 1874 optional uCDN-internal process, this does not require any 1875 standardization effort beyond being able to communicate chunk 1876 locations in the CDNI Request Routing Redirection interface. 1878 Third, with respect to the CDNI Logging interface, there are several 1879 potential issues, including the large number of individual chunk 1880 requests potentially resulting in large volumes of logging 1881 information, and the desire to correlate logging information for 1882 chunk requests that correspond to the same HAS session. For the 1883 initial CDNI specification, our approach is to expect participating 1884 CDNs to support per-chunk logging (e.g. logging each chunk request as 1885 if it were an independent content request) over the CDNI Logging 1886 interface. Optionally, the LI may include a Content Collection 1887 IDentifier (CCID) and/or a Session IDentifier (SID) as part of the 1888 logging fields, thereby facilitating correlation of per-chunk logs 1889 into per-session logs for applications benefiting from such session 1890 level information (e.g. session-based analytics). This approach is 1891 in line with the recommendations made in RFC 6983, which also 1892 identifies potential improvements in this area that might be 1893 considered in the future. 1895 Fourth, with respect to the CDNI Control interface, and in particular 1896 purging HAS chunks from a given CDN, our approach is to expect each 1897 CDN supports per-chunk content purge (e.g. purging of chunks as if 1898 they were individual content items). Optionally, a CDN may support 1899 content purge on the basis of a "Purge IDentifier (Purge-ID)" 1900 allowing the removal of all chunks related to a given Content 1901 Collection with a single reference. It is possible that this Purge- 1902 ID could be merged with the CCID discussed above for HAS Logging, or 1903 alternatively, they may remain distinct. 1905 4.8. URI Rewriting 1907 When using HTTP redirection, content URIs may be rewritten when 1908 redirection takes place within an uCDN, from an uCDN to a dCDN, and 1909 within the dCDN. In the case of cascaded CDNs, content URIs may be 1910 rewritten at every CDN hop (e.g., between the uCDN and the dCDN 1911 acting as the transit CDN, and between the transit CDN and the dCDN 1912 serving the request. The content URI used between any uCDN/dCDN pair 1913 becomes a common handle that can be referred without ambiguity by 1914 both CDNs in all their inter-CDN communications. This handle allows 1915 the uCDN and dCDN to correlate information exchanged using other CDNI 1916 interfaces in both the downstream direction (e.g., when using the MI) 1917 and the upstream direction (e.g., when using the LI). 1919 Consider the simple case of a single uCDN/dCDN pair using HTTP 1920 redirection. We introduce the following terminology for content URIs 1921 to simplify the discussion: 1923 "u-URI" represents a content URI in a request presented to the 1924 uCDN; 1926 "ud-URI" is a content URI acting as the common handle across uCDN 1927 and dCDN for requests redirected by the uCDN to a specific dCDN; 1929 "d-URI" represents a content URI in a request made within the 1930 delegate dCDN. 1932 In our simple pair-wise example, the "ud-URI" effectively becomes the 1933 handle that the uCDN/dCDN pair use to correlate all CDNI information. 1934 In particular, for a given pair of CDNs executing the HTTP 1935 redirection, the uCDN needs to map the u-URI to the ud-URI handle for 1936 all MI message exchanges, while the dCDN needs to map the d-URI to 1937 the ud-URI handle for all LI message exchanges. 1939 In the case of a cascaded CDNs, the transit CDN will rewrite the 1940 content URI when redirecting to the dCDN, thereby establishing a new 1941 handle between the transit CDN and the dCDN, that is different from 1942 the handle between the uCDN and transit CDN. It is the 1943 responsibility of the transit CDN to manage its mapping across 1944 handles so the right handle for all pairs of CDNs is always used in 1945 its CDNI communication. 1947 In summary, all CDNI interfaces between a given pair of CDNs need to 1948 always use the "ud-URI" handle for that specific CDN pair as their 1949 content URI reference. 1951 5. Deployment Models 1953 In this section we describe a number of possible deployment models 1954 that may be achieved using the CDNI interfaces described above. We 1955 note that these models are by no means exhaustive, and that many 1956 other models may be possible. 1958 Although the reference model of Figure 1 shows all CDN functions on 1959 each side of the CDNI interface, deployments can rely on entities 1960 that are involved in any subset of these functions, and therefore 1961 only support the relevant subset of CDNI interfaces. As already 1962 noted in Section 3, effective CDNI deployments can be built without 1963 necessarily implementing all the interfaces. Some examples of such 1964 deployments are shown below. 1966 Note that, while we refer to upstream and downstream CDNs, this 1967 distinction applies to specific content items and transactions. That 1968 is, a given CDN may be upstream for some transactions and downstream 1969 for others, depending on many factors such as location of the 1970 requesting client and the particular piece of content requested. 1972 5.1. Meshed CDNs 1974 Although the reference model illustrated in Figure 1 shows a 1975 unidirectional CDN interconnection with a single uCDN and a single 1976 dCDN, any arbitrary CDNI meshing can be built from this, such as the 1977 example meshing illustrated in Figure 11. (Support for arbitrary 1978 meshing may or may not be in the initial scope for the working group, 1979 but the model allows for it.) 1980 ------------- ----------- 1981 / CDN A \<==CDNI===>/ CDN B \ 1982 \ / \ / 1983 ------------- ----------- 1984 /\ \\ /\ 1985 || \\ || 1986 CDNI \==CDNI===\\ CDNI 1987 || \\ || 1988 \/ \/ \/ 1989 ------------- ----------- 1990 / CDN C \===CDNI===>/ CDN D \ 1991 \ / \ / 1992 ------------- ----------- 1993 /\ 1994 || 1995 CDNI 1996 || 1997 \/ 1998 ------------- 1999 / CDN E \ 2000 \ / 2001 ------------- 2003 ===> CDNI interfaces, with right-hand side CDN acting as dCDN 2004 to left-hand side CDN 2005 <==> CDNI interfaces, with right-hand side CDN acting as dCDN 2006 to left-hand side CDN and with left-hand side CDN acting 2007 as dCDN to right-hand side CDN 2009 Figure 11: CDNI Deployment Model: CDN Meshing Example 2011 5.2. CSP combined with CDN 2013 Note that our terminology refers to functional roles and not economic 2014 or business roles. That is, a given organization may be operating as 2015 both a CSP and a fully fledged uCDN when we consider the functions 2016 performed, as illustrated in Figure 12. 2018 ##################################### ################## 2019 # # # # 2020 # Organization A # # Organization B # 2021 # # # # 2022 # -------- ------------- # # ----------- # 2023 # / CSP \ / uCDN \ # # / dCDN \ # 2024 # | | | +----+ | # # | +----+ | # 2025 # | | | | C | | # # | | C | | # 2026 # | | | +----+ | # # | +----+ | # 2027 # | | | +----+ | # # | +----+ | # 2028 # | | | | L | | # # | | L | | # 2029 # | |*****| +----+ |===CDNI===>| +----+ | # 2030 # | | | +----+ | # # | +----+ | # 2031 # | | | | RR | | # # | | RR | | # 2032 # | | | +----+ | # # | +----+ | # 2033 # | | | +----+ | # # | +----+ | # 2034 # | | | | D | | # # | | D | | # 2035 # | | | +----+ | # # | +----+ | # 2036 # \ / \ / # # \ / # 2037 # -------- ------------- # # ----------- # 2038 # # # # 2039 ##################################### ################## 2041 ===> CDNI interfaces, with right-hand side CDN acting as dCDN 2042 to left-hand side CDN 2043 **** interfaces outside the scope of CDNI 2044 C Control component of the CDN 2045 L Logging component of the CDN 2046 RR Request Routing component of the CDN 2047 D Distribution component of the CDN 2049 Figure 12: CDNI Deployment Model: Organization combining CSP & uCDN 2051 5.3. CSP using CDNI Request Routing Interface 2053 As another example, a content provider organization may choose to run 2054 its own request routing function as a way to select among multiple 2055 candidate CDN providers; In this case the content provider may be 2056 modeled as the combination of a CSP and of a special, restricted case 2057 of a CDN. In that case, as illustrated in Figure 13, the CDNI 2058 Request Routing interfaces can be used between the restricted CDN 2059 operated by the content provider Organization and the CDN operated by 2060 the full CDN organization acting as a dCDN in the request routing 2061 control plane. Interfaces outside the scope of the CDNI work can be 2062 used between the CSP functional entities of the content provider 2063 organization and the CDN operated by the full CDN organization acting 2064 as a uCDN) in the CDNI control planes other than the request routing 2065 plane (i.e. Control, Distribution, Logging). 2067 ##################################### ################## 2068 # # # # 2069 # Organization A # # Organization B # 2070 # # # # 2071 # -------- ------------- # # ----------- # 2072 # / CSP \ / uCDN(RR) \ # # / dCDN(RR) \ # 2073 # | | | +----+ | # # | +----+ | # 2074 # | |*****| | RR |==========CDNI=====>| RR | | # 2075 # | | | +----+ | # RR # | +----+ | # 2076 # | | \ / # # | | # 2077 # | | ------------- # # |uCDN(C,L,D)| # 2078 # | | # # | +----+ | # 2079 # | | # # | | C | | # 2080 # | |*******************************| +----+ | # 2081 # | | # # | +----+ | # 2082 # | | # # | | L | | # 2083 # | | # # | +----+ | # 2084 # | | # # | +----+ | # 2085 # | | # # | | D | | # 2086 # | | # # | +----+ | # 2087 # \ / # # \ / # 2088 # -------- # # ----------- # 2089 # # # # 2090 ##################################### ################## 2092 ===> CDNI Request Routing Interface 2093 **** interfaces outside the scope of CDNI 2095 Figure 13: CDNI Deployment Model: Organization combining CSP and 2096 partial CDN 2098 5.4. CDN Federations and CDN Exchanges 2100 There are two additional concepts related to, but distinct from CDN 2101 Interconnection. The first is CDN Federation. Our view is that CDNI 2102 is the more general concept, involving two or more CDNs serving 2103 content to each other's users, while federation implies a multi- 2104 lateral interconnection arrangement, but other CDN interconnection 2105 agreements are also possible (e.g., symmetric bilateral, asymmetric 2106 bilateral). An important conclusion is that CDNI technology should 2107 not presume (or bake in) a particular interconnection agreement, but 2108 should instead be general enough to permit alternative 2109 interconnection arrangements to evolve. 2111 The second concept often used in the context of CDN Federation is CDN 2112 Exchange--a third party broker or exchange that is used to facilitate 2113 a CDN federation. Our view is that a CDN exchange offers valuable 2114 machinery to scale the number of CDN operators involved in a multi- 2115 lateral (federated) agreement, but that this machinery is built on 2116 top of the core CDNI interconnection mechanisms. For example, as 2117 illustrated in Figure 14, the exchange might aggregate and 2118 redistribute information about each CDN footprint and capacity, as 2119 well as collect, filter, and redistribute traffic logs that each 2120 participant needs for interconnection settlement, but inter-CDN 2121 request routing, inter-CDN content distribution (including inter-CDN 2122 acquisition) and inter-CDN control which fundamentally involve a 2123 direct interaction between an upstream CDN and a downstream CDN-- 2124 operate exactly as in a pair-wise peering arrangement. Turning to 2125 Figure 14, we observe that in this example: 2127 o each CDN supports a direct CDNI Control interface to every other 2128 CDN 2130 o each CDN supports a direct CDNI Metadata interface to every other 2131 CDN 2133 o each CDN supports a CDNI Logging interface with the CDN Exchange 2135 o each CDN supports both a CDNI Request Routing interfaces with the 2136 CDN Exchange (for aggregation and redistribution of dynamic CDN 2137 footprint discovery information) and a direct RI to every other 2138 CDN (for actual request redirection). 2140 ---------- --------- 2141 / CDN A \ / CDN B \ 2142 | +----+ | | +----+ | 2143 //========>| C |<==============CDNI============>| C |<==========\\ 2144 || | +----+ | C | +----+ | || 2145 || | +----+ | | +----+ | || 2146 || //=====>| D |<==============CDNI============>| D |<=======\\ || 2147 || || | +----+ | M | +----+ | || || 2148 || || | | /------------\ | | || || 2149 || || | +----+ | | +--+ CDN Ex| | +----+ | || || 2150 || || //==>| RR |<===CDNI==>|RR|<=======CDNI====>| RR |<====\\ || || 2151 || || || | +----+ | RR | +--+ | RR | +----+ | || || || 2152 || || || | | | /\ | | | || || || 2153 || || || | +----+ | | || +---+ | | +----+ | || || || 2154 || || || | | L |<===CDNI=======>| L |<=CDNI====>| L | | || || || 2155 || || || | +----+ | L | || +---+ | L | +----+ | || || || 2156 || || || \ / \ || /\ / \ / || || || 2157 || || || ----------- --||----||-- ----------- || || || 2158 || || || || || || || || 2159 || || || CDNI RR || || || || 2160 || || || || CDNI L || || || 2161 || || || || || || || || 2162 || || || ---||----||---- || || || 2163 || || || / \/ || \ || || || 2164 || || || | +----+ || | || || || 2165 || || \\=====CDNI==========>| RR |<=============CDNI========// || || 2166 || || RR | +----+ \/ | RR || || 2167 || || | +----+ | || || 2168 || || | | L | | || || 2169 || || | +----+ | || || 2170 || || | +----+ | || || 2171 || \\=======CDNI===========>| D |<=============CDNI===========// || 2172 || M | +----+ | M || 2173 || | +----+ | || 2174 \\==========CDNI===========>| C |<=============CDNI==============// 2175 C | +----+ | C 2176 \ CDN C / 2177 -------------- 2179 <=CDNI RR=> CDNI Request Routing Interface 2180 <=CDNI M==> CDNI Metadata Interface 2181 <=CDNI C==> CDNI Control Interface 2182 <=CDNI L==> CDNI Logging Interface 2184 Figure 14: CDNI Deployment Model: CDN Exchange 2186 Note that a CDN exchange may alternatively support a different set of 2187 functionality (e.g. Logging only, or Logging and full request 2188 routing, or all the functionality of a CDN including content 2189 distribution). All these options are expected to be allowed by the 2190 IETF CDNI specifications. 2192 6. Trust Model 2194 There are a number of trust issues that need to be addressed by a 2195 CDNI solution. Many of them are in fact similar or identical to 2196 those in a simple CDN without interconnection. In a standard CDN 2197 environment (without CDNI), the CSP places a degree of trust in a 2198 single CDN operator to perform many functions. The CDN is trusted to 2199 deliver content with appropriate quality of experience for the end 2200 user. The CSP trusts the CDN operator not to corrupt or modify the 2201 content. The CSP often relies on the CDN operator to provide 2202 reliable accounting information regarding the volume of delivered 2203 content. The CSP may also trust the CDN operator to perform actions 2204 such as timely invalidation of content and restriction of access to 2205 content based on certain criteria such as location of the user and 2206 time of day, and to enforce per-request authorization performed by 2207 the CSP using techniques such as URI signing. 2209 A CSP also places trust in the CDN not to distribute any information 2210 that is confidential to the CSP (e.g., how popular a given piece of 2211 content is) or confidential to the end user (e.g., which content has 2212 been watched by which user). 2214 A CSP does not necessarily have to place complete trust in a CDN. A 2215 CSP will in some cases take steps to protect its content from 2216 improper distribution by a CDN, e.g. by encrypting it and 2217 distributing keys in some out of band way. A CSP also depends on 2218 monitoring (possibly by third parties) and reporting to verify that 2219 the CDN has performed adequately. A CSP may use techniques such as 2220 client-based metering to verify that accounting information provided 2221 by the CDN is reliable. HTTP conditional requests may be used to 2222 provide the CSP with some checks on CDN operation. In other words, 2223 while a CSP may trust a CDN to perform some functions in the short 2224 term, the CSP is able in most cases to verify whether these actions 2225 have been performed correctly and to take action (such as moving the 2226 content to a different CDN) if the CDN does not live up to 2227 expectations. 2229 The main trust issue raised by CDNI is that it introduces transitive 2230 trust. A CDN that has a direct relationship with a CSP can now 2231 "outsource" the delivery of content to another (downstream) CDN. 2232 That CDN may in term outsource delivery to yet another downstream 2233 CDN, and so on. 2235 The top level CDN in such a chain of delegation is responsible for 2236 ensuring that the requirements of the CSP are met. Failure to do so 2237 is presumably just as serious as in the traditional single CDN case. 2238 Hence, an upstream CDN is essentially trusting a downstream CDN to 2239 perform functions on its behalf in just the same way as a CSP trusts 2240 a single CDN. Monitoring and reporting can similarly be used to 2241 verify that the downstream CDN has performed appropriately. However, 2242 the introduction of multiple CDNs in the path between CSP and end 2243 user complicates the picture. For example, third party monitoring of 2244 CDN performance (or other aspects of operation, such as timely 2245 invalidation) might be able to identify the fact that a problem 2246 occurred somewhere in the chain but not point to the particular CDN 2247 at fault. 2249 In summary, we assume that an upstream CDN will invest a certain 2250 amount of trust in a downstream CDN, but that it will verify that the 2251 downstream CDN is performing correctly, and take corrective action 2252 (including potentially breaking off its relationship with that CDN) 2253 if behavior is not correct. We do not expect that the trust 2254 relationship between a CSP and its "top level" CDN will differ 2255 significantly from that found today in single CDN situations. 2256 However, it does appear that more sophisticated tools and techniques 2257 for monitoring CDN performance and behavior will be required to 2258 enable the identification of the CDN at fault in a particular 2259 delivery chain. 2261 We expect that the detailed designs for the specific interfaces for 2262 CDNI will need to take the transitive trust issues into account. For 2263 example, explicit confirmation that some action (such as content 2264 removal) has taken place in a downstream CDN may help to mitigate 2265 some issues of transitive trust. 2267 7. IANA Considerations 2269 This memo includes no request to IANA. 2271 8. Security Considerations 2273 While there is a variety of security issues introduced by a single 2274 CDN, we are concerned here specifically with the additional issues 2275 that arise when CDNs are interconnected. For example, when a single 2276 CDN has the ability to distribute content on behalf of a CSP, there 2277 may be concerns that such content could be distributed to parties who 2278 are not authorized to receive it, and there are mechanisms to deal 2279 with such concerns. Our focus in this section is on how CDN 2280 interconnection introduces new security issues not found in the 2281 single CDN case. 2283 Many of the security issues that arise in CDNI are related to the 2284 transitivity of trust (or lack thereof) described in Section 6. As 2285 noted above, the design of the various interfaces for CDNI must take 2286 account of the additional risks posed by the fact that a CDN with 2287 whom a CSP has no direct relationship is now potentially distributing 2288 content for that CSP. The mechanisms used to mitigate these risks 2289 may be similar to those used in the single CDN case, but their 2290 suitability in this more complex environment must be validated. 2292 Another concern that arises in any CDN is that information about the 2293 behavior of users (what content they access, how much content they 2294 consume, etc.) may be gathered by the CDN. This risk certainly 2295 exists in inter-connected CDNs, but it should be possible to apply 2296 the same techniques to mitigate it as in the single CDN case. 2298 CDNs today offer a variety of means to control access to content, 2299 such as time-of-day restrictions, geo-blocking, and URI signing. 2300 These mechanisms must continue to function in CDNI environments, and 2301 this consideration is likely to affect the design of certain CDNI 2302 interfaces (e.g. metadata, request routing.) 2304 Just as with a single CDN, each peer CDN must ensure that it is not 2305 used as an "open proxy" to deliver content on behalf of a malicious 2306 CSP. Whereas a single CDN typically addresses this problem by having 2307 CSPs explicitly register content (or origin servers) that is to be 2308 served, simply propagating this information to peer downstream CDNs 2309 may be problematic because it reveals more information than the 2310 upstream CDN is willing to specify. (To this end, the content 2311 acquisition step in the earlier examples force the dCDN to retrieve 2312 content from the uCDN rather than go directly to the origin server.) 2314 There are several approaches to this problem. One is for the uCDN to 2315 encoded a signed token generated from a shared secret in each URL 2316 routed to a dCDN, and for the dCDN to validate the request based on 2317 this token. Another one is to have each upstream CDN advertise the 2318 set of CDN-Domains they serve, where the downstream CDN checks each 2319 request against this set before caching and delivering the associated 2320 object. Although straightforward, this approach requires operators 2321 to reveal additional information, which may or may not be an issue. 2323 8.1. Security of CDNI Interfaces 2325 It is noted in [I-D.ietf-cdni-requirements] that all CDNI interfaces 2326 must be able to operate securely over insecure IP networks. Since it 2327 is expected that the CDNI interfaces will be implemented using 2328 existing application protocols such as HTTP or XMPP, we also expect 2329 that the security mechanisms available to those protocols may be used 2330 by the CDNI interfaces. Details of how these interfaces are secured 2331 will be specified in the relevant interface documents. 2333 8.2. Digital Rights Management 2335 Issues of digital rights management (DRM, also sometimes called 2336 digital restrictions management) is often employed for content 2337 distributed via CDNs. In general, DRM relies on the CDN to 2338 distribute encrypted content, with decryption keys distributed to 2339 users by some other means (e.g. directly from the CSP to the end 2340 user.) For this reason, DRM is considered out of scope for the CDNI 2341 WG [RFC6707] and does not introduce additional security issues for 2342 CDNI. 2344 9. Contributors 2346 The following individuals contributed to this document: 2348 o Ray van Brandenburg 2350 o Matt Caulfield 2352 o Francois le Faucheur 2354 o Aaron Falk 2356 o David Ferguson 2358 o John Hartman 2360 o Ben Niven-Jenkins 2362 o Kent Leung 2364 10. Acknowledgements 2366 We thank Huw Jones for helpful input to the draft. 2368 11. Informative References 2370 [I-D.ietf-appsawg-http-forwarded] 2371 Petersson, A. and M. Nilsson, "Forwarded HTTP Extension", 2372 draft-ietf-appsawg-http-forwarded-10 (work in progress), 2373 October 2012. 2375 [I-D.ietf-cdni-control-triggers] 2376 Murray, R. and B. Niven-Jenkins, "CDNI Control Interface / 2377 Triggers", draft-ietf-cdni-control-triggers-02 (work in 2378 progress), December 2013. 2380 [I-D.ietf-cdni-footprint-capabilities-semantics] 2381 Seedorf, J., Peterson, J., Previdi, S., Brandenburg, R., 2382 and K. Ma, "CDNI Request Routing: Footprint and 2383 Capabilities Semantics", draft-ietf-cdni-footprint- 2384 capabilities-semantics-01 (work in progress), October 2385 2013. 2387 [I-D.ietf-cdni-logging] 2388 Faucheur, F., Bertrand, G., Oprescu, I., and R. 2389 Peterkofsky, "CDNI Logging Interface", draft-ietf-cdni- 2390 logging-08 (work in progress), October 2013. 2392 [I-D.ietf-cdni-metadata] 2393 Niven-Jenkins, B., Murray, R., Watson, G., Caulfield, M., 2394 Leung, K., and K. Ma, "CDN Interconnect Metadata", draft- 2395 ietf-cdni-metadata-04 (work in progress), December 2013. 2397 [I-D.ietf-cdni-redirection] 2398 Danhua, W., Niven-Jenkins, B., He, X., Chen, G., and W. 2399 Ni, "Request Routing Redirection Interface for CDN 2400 Interconnection", draft-ietf-cdni-redirection-01 (work in 2401 progress), October 2013. 2403 [I-D.ietf-cdni-requirements] 2404 Leung, K. and Y. Lee, "Content Distribution Network 2405 Interconnection (CDNI) Requirements", draft-ietf-cdni- 2406 requirements-14 (work in progress), December 2013. 2408 [I-D.vandergaast-edns-client-subnet] 2409 Contavalli, C., Gaast, W., Leach, S., and E. Lewis, 2410 "Client Subnet in DNS Requests", draft-vandergaast-edns- 2411 client-subnet-02 (work in progress), July 2013. 2413 [RFC3466] Day, M., Cain, B., Tomlinson, G., and P. Rzewski, "A Model 2414 for Content Internetworking (CDI)", RFC 3466, February 2415 2003. 2417 [RFC6707] Niven-Jenkins, B., Le Faucheur, F., and N. Bitar, "Content 2418 Distribution Network Interconnection (CDNI) Problem 2419 Statement", RFC 6707, September 2012. 2421 [RFC6770] Bertrand, G., Stephan, E., Burbridge, T., Eardley, P., Ma, 2422 K., and G. Watson, "Use Cases for Content Delivery Network 2423 Interconnection", RFC 6770, November 2012. 2425 [RFC6983] van Brandenburg, R., van Deventer, O., Le Faucheur, F., 2426 and K. Leung, "Models for HTTP-Adaptive-Streaming-Aware 2427 Content Distribution Network Interconnection (CDNI)", RFC 2428 6983, July 2013. 2430 Authors' Addresses 2432 Larry Peterson (editor) 2433 Akamai Technologies, Inc. 2434 8 Cambridge Center 2435 Cambridge, MA 02142 2436 USA 2438 Email: lapeters@akamai.com 2440 Bruce Davie 2441 VMware, Inc. 2442 3401 Hillview Ave. 2443 Palo Alto, CA 94304 2444 USA 2446 Email: bdavie@vmware.com