idnits 2.17.1 draft-ietf-forces-interfelfb-01.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (March 6, 2015) is 3339 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Downref: Normative reference to an Informational RFC: RFC 3746 Summary: 1 error (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Internet Engineering Task Force D. Joachimpillai 3 Internet-Draft Verizon 4 Intended status: Standards Track J. Hadi Salim 5 Expires: September 7, 2015 Mojatatu Networks 6 March 6, 2015 8 ForCES Inter-FE LFB 9 draft-ietf-forces-interfelfb-01 11 Abstract 13 This document describes extending the ForCES LFB topology across FEs 14 i.e inter-FE connectivity without needing any changes to the ForCES 15 specification by defining the Inter-FE LFB. The Inter-FE LFB 16 provides ability to pass data, metadata and exceptions across FEs. 17 The document describes a generic way to transport the mentioned 18 details but focuses on ethernet transport. 20 Status of this Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on September 7, 2015. 37 Copyright Notice 39 Copyright (c) 2015 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Terminology and Conventions . . . . . . . . . . . . . . . . . 3 55 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 56 1.2. Definitions . . . . . . . . . . . . . . . . . . . . . . . 3 57 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 58 3. Problem Scope And Use Cases . . . . . . . . . . . . . . . . . 4 59 3.1. Basic Router . . . . . . . . . . . . . . . . . . . . . . . 4 60 3.1.1. Distributing The LFB Topology . . . . . . . . . . . . 6 61 3.2. Arbitrary Network Function . . . . . . . . . . . . . . . . 7 62 3.2.1. Distributing The Arbitrary Network Function . . . . . 8 63 4. Proposal Overview . . . . . . . . . . . . . . . . . . . . . . 9 64 4.1. Inserting The Inter-FE LFB . . . . . . . . . . . . . . . . 9 65 5. Generic Inter-FE connectivity . . . . . . . . . . . . . . . . 11 66 5.1. Inter-FE Ethernet Connectivity . . . . . . . . . . . . . . 13 67 5.1.1. Inter-FE Ethernet Connectivity Issues . . . . . . . . 15 68 6. Detailed Description of the Ethernet inter-FE LFB . . . . . . 16 69 6.1. Data Handling . . . . . . . . . . . . . . . . . . . . . . 16 70 6.1.1. Egress Processing . . . . . . . . . . . . . . . . . . 17 71 6.1.2. Ingress Processing . . . . . . . . . . . . . . . . . . 18 72 6.2. Components . . . . . . . . . . . . . . . . . . . . . . . . 19 73 6.3. Inter-FE LFB XML Model . . . . . . . . . . . . . . . . . . 19 74 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 24 75 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 24 76 9. IEEE Assignment Considerations . . . . . . . . . . . . . . . . 24 77 10. Security Considerations . . . . . . . . . . . . . . . . . . . 24 78 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 25 79 11.1. Normative References . . . . . . . . . . . . . . . . . . . 25 80 11.2. Informative References . . . . . . . . . . . . . . . . . . 25 81 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 26 83 1. Terminology and Conventions 85 1.1. Requirements Language 87 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 88 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 89 document are to be interpreted as described in [RFC2119]. 91 1.2. Definitions 93 This document reiterates the terminology defined in several ForCES 94 documents [RFC3746], [RFC5810], [RFC5811], and [RFC5812] for the sake 95 of contextual clarity. 97 Control Engine (CE) 99 Forwarding Engine (FE) 101 FE Model 103 LFB (Logical Functional Block) Class (or type) 105 LFB Instance 107 LFB Model 109 LFB Metadata 111 ForCES Component 113 LFB Component 115 ForCES Protocol Layer (ForCES PL) 117 ForCES Protocol Transport Mapping Layer (ForCES TML) 119 2. Introduction 121 In the ForCES architecture, a packet service can be modelled by 122 composing a graph of one or more LFB instances. The reader is 123 referred to the details in the ForCES Model [RFC5812]. 125 The FEObject LFB capabilities in the ForCES Model [RFC5812] define 126 component ModifiableLFBTopology which, when advertised by the FE, 127 implies that the advertising FE is capable of allowing creation and 128 modification of LFB graph(s) by the control plane. Details on how a 129 graph of LFB class instances can be created can be derived by the 130 control plane by looking at the FE's FEObject LFB class table 131 component SupportedLFBs. The SupportedLFBs table contains 132 information about each LFB class that the FE supports. For each LFB 133 class supported, details are provided on how the supported LFB class 134 may be connected to other LFB classes. The SupportedLFBs table 135 describes which LFB class a specified LFB class may succeed or 136 precede in an LFB class instance topology. Each link connecting two 137 LFB class instances is described in the LFBLinkType dataTypeDef and 138 has sufficient details to identify precisely the end points of a link 139 of a service graph. 141 The CE may therefore create a packet service by describing an LFB 142 instance graph connection; this is achieved by updating the FEOBject 143 LFBTopology table. 145 Often there are requirements for the packet service graph to cross FE 146 boundaries. This could be from a desire to scale the service or need 147 to interact with LFBs which reside in a separate FE (eg lookaside 148 interface to a shared TCAM, an interconnected chip, or as coarse 149 grained functionality as an external NAT FE box being part of the 150 service graph etc). 152 Given that the ForCES inter-LFB architecture calls out for ability to 153 pass metadata between LFBs, it is imperative therefore to define 154 mechanisms to extend that existing feature and allow passing the 155 metadata between LFBs across FEs. 157 This document describes extending the LFB topology across FEs i.e 158 inter-FE connectivity without needing any changes to the ForCES 159 definitions. It focusses on using Ethernet as the interconnection as 160 a starting point while leaving room for other protocols (such as 161 directly on top of IP, UDP, VXLAN, etc) to be addressed by other 162 future documents. 164 3. Problem Scope And Use Cases 166 The scope of this document is to solve the challenge of passing 167 ForCES defined metadata and exceptions across FEs (be they physical 168 or virtual). To illustrate the problem scope we present two use 169 cases where we start with a single FE running all the functionality 170 then split it into multiple FEs. 172 3.1. Basic Router 174 A sample LFB topology Figure 1 demonstrates a service graph for 175 delivering basic IPV4 forwarding service within one FE. For the 176 purpose of illustration, the diagram shows LFB classes as graph nodes 177 instead of multiple LFB class instances. 179 Since the illustration is meant only as an exercise to showcase how 180 data and metadata are sent down or upstream on a graph of LFBs, it 181 abstracts out any ports in both directions and talks about a generic 182 ingress and egress LFB. Again, for illustration purposes, the 183 diagram does not show exception or error paths. Also left out are 184 details on Reverse Path Filtering, ECMP, multicast handling etc. In 185 other words, this is not meant to be a complete description of an 186 IPV4 forwarding application; for a more complete example, please 187 refer to the LFBlib document [RFC6956]. 189 The output of the ingress LFB(s) coming into the IPv4 Validator LFB 190 will have both the IPV4 packets and, depending on the implementation, 191 a variety of ingress metadata such as offsets into the different 192 headers, any classification metadata, physical and virtual ports 193 encountered, tunnelling information etc. These metadata are lumped 194 together as "ingress metadata". 196 Once the IPV4 validator vets the packet (example ensures that no 197 expired TTL etc), it feeds the packet and inherited metadata into the 198 IPV4 unicast LPM LFB. 200 +----+ 201 | | 202 IPV4 pkt | | IPV4 pkt +-----+ +---+ 203 +------------->| +------------->| | | | 204 | + ingress | | + ingress |IPv4 | IPV4 pkt | | 205 | metadata | | metadata |Ucast+------------>| +--+ 206 | +----+ |LPM | + ingress | | | 207 +-+-+ IPv4 +-----+ + NHinfo +---+ | 208 | | Validator metadata IPv4 | 209 | | LFB NextHop| 210 | | LFB | 211 | | | 212 | | IPV4 pkt | 213 | | + {ingress | 214 +---+ + NHdetails} 215 Ingress metadata | 216 LFB +--------+ | 217 | Egress | | 218 <--+ |<-----------------+ 219 | LFB | 220 +--------+ 222 Figure 1: Basic IPV4 packet service LFB topology 224 The IPV4 unicast LPM LFB does a longest prefix match lookup on the 225 IPV4 FIB using the destination IP address as a search key. The 226 result is typically a next hop selector which is passed downstream as 227 metadata. 229 The Nexthop LFB receives the IPv4 packet with an associated next hop 230 info metadata. The NextHop LFB consumes the NH info metadata and 231 derives from it a table index to look up the next hop table in order 232 to find the appropriate egress information. The lookup result is 233 used to build the next hop details to be used downstream on the 234 egress. This information may include any source and destination 235 information (MAC address to use, if ethernet;) as well egress ports. 236 [Note: It is also at this LFB where typically the forwarding TTL 237 decrement and IP checksum recalculation occurs.] 239 The details of the egress LFB are considered out of scope for this 240 discussion. Suffice it is to say that somewhere within or beyond the 241 Egress LFB the IPV4 packet will be sent out a port (ethernet, virtual 242 or physical etc). 244 3.1.1. Distributing The LFB Topology 246 Figure 2 demonstrates one way the router LFB topology in Figure 1 may 247 be split across two FEs (eg two ASICs). Figure 2 shows the LFB 248 topology split across FEs after the IPV4 unicast LPM LFB. 250 FE1 251 +-------------------------------------------------------------+ 252 | +----+ | 253 | +----------+ | | | 254 | | Ingress | IPV4 pkt | | IPV4 pkt +-----+ | 255 | | LFB +-------------->| +------------->| | | 256 | | | + ingress | | + ingress |IPv4 | | 257 | +----------+ metadata | | metadata |Ucast| | 258 | ^ +----+ |LPM | | 259 | | IPv4 +--+--+ | 260 | | Validator | | 261 | LFB | | 262 +---------------------------------------------------|---------+ 263 | 264 IPv4 packet + 265 {ingress + NHinfo} 266 metadata 267 FE2 | 268 +---------------------------------------------------|---------+ 269 | V | 270 | +--------+ +--------+ | 271 | | Egress | IPV4 packet | IPV4 | | 272 | <-----+ LFB |<----------------------+NextHop | | 273 | | |{ingress + NHdetails} | LFB | | 274 | +--------+ metadata +--------+ | 275 +-------------------------------------------------------------+ 277 Figure 2: Split IPV4 packet service LFB topology 279 Some proprietary inter-connect (example Broadcom Higig over XAUI 280 [brcm-higig]) are known to exist to carry both the IPV4 packet and 281 the related metadata between the IPV4 Unicast LFB and IPV4 NextHop 282 LFB across the two FEs. 284 The purpose of the inter-FE LFB is to define standard mechanisms for 285 interconnecting FEs and for that reason we are not going to touch 286 anymore on proprietary chip-chip interconnects other than state the 287 fact they exist and that it is feasible to have translation to and 288 from proprietary approaches. The document focus is the FE-FE 289 interconnect where the FE could be physical or virtual and the 290 interconnecting technology runs a standard protocol such as ethernet, 291 IP or other protocols on top of IP. 293 3.2. Arbitrary Network Function 295 In this section we show an example of an arbitrary network function 296 which is more coarse grained in terms of functionality. Each Network 297 function may constitute more than one LFB. 299 FE1 300 +-------------------------------------------------------------+ 301 | +----+ | 302 | +----------+ | | | 303 | | Network | pkt |NF2 | pkt +-----+ | 304 | | Function +-------------->| +------------->| | | 305 | | 1 | + NF1 | | + NF1/2 |NF3 | | 306 | +----------+ metadata | | metadata | | | 307 | ^ +----+ | | | 308 | | +--+--+ | 309 | | | | 310 | | | 311 +---------------------------------------------------|---------+ 312 V 314 Figure 3: A Network Function Service Chain within one FE 316 The setup in Figure 3 is a typical of most packet processing boxes 317 where we have functions like DPI, NAT, Routing, etc connected in such 318 a topology to deliver a packet processing service to flows. 320 3.2.1. Distributing The Arbitrary Network Function 322 The setup in Figure 3 can be split out across 3 FEs instead as 323 demonstrated in Figure 4. This could be motivated by scale out 324 reasons or because different vendors provide different functionality 325 which is plugged-in to provide such functionality. The end result is 326 to have the same packet service delivered to the different flows 327 passing through. 329 FE1 FE2 330 +----------+ +----+ FE3 331 | Network | pkt |NF2 | pkt +-----+ 332 | Function +-------------->| +------------->| | 333 | 1 | + NF1 | | + NF1/2 |NF3 | 334 +----------+ metadata | | metadata | | 335 ^ +----+ | | 336 | +--+--+ 337 | 338 V 340 Figure 4: A Network Function Service Chain Distributed Across 341 Multiple FEs 343 4. Proposal Overview 345 We address the inter-FE connectivity requirements by proposing the 346 inter-FE LFB class. Using a standard LFB class definition implies no 347 change to the basic ForCES architecture in the form of the core LFBs 348 (FE Protocol or Object LFBs). This design choice was made after 349 considering an alternative approach that would have required changes 350 to both the FE Object capabilities (SupportedLFBs) as well 351 LFBTopology component to describe the inter-FE connectivity 352 capabilities as well as runtime topology of the LFB instances. 354 4.1. Inserting The Inter-FE LFB 356 The distributed LFB topology described in Figure 2 is re-illustrated 357 in Figure 5 to show the topology location where the inter-FE LFB 358 would fit in. 360 FE1 361 +-------------------------------------------------------------+ 362 | +----------+ +----+ | 363 | | Ingress | IPV4 pkt | | IPV4 pkt +-----+ | 364 | | LFB +-------------->| +------------->| | | 365 | | | + ingress | | + ingress |IPv4 | | 366 | +----------+ metadata | | metadata |Ucast| | 367 | ^ +----+ |LPM | | 368 | | IPv4 +--+--+ | 369 | | Validator | | 370 | | LFB | | 371 | | IPv4 pkt + metadata | 372 | | {ingress + NHinfo + InterFEid}| 373 | | | | 374 | +----V----+ | 375 | | InterFE | | 376 | | LFB | | 377 | +----+----+ | 378 +---------------------------------------------------|---------+ 379 | 380 IPv4 packet and metadata 381 {ingress + NHinfo + Inter FE info} 382 FE2 | 383 +---------------------------------------------------|---------+ 384 | +----V----+ | 385 | | InterFE | | 386 | | LFB | | 387 | +----+----+ | 388 | | | 389 | IPv4 pkt + metadata | 390 | {ingress + NHinfo} | 391 | | | 392 | +--------+ +----V---+ | 393 | | Egress | IPV4 packet | IPV4 | | 394 | <-----+ LFB |<----------------------+NextHop | | 395 | | |{ingress + NHdetails} | LFB | | 396 | +--------+ metadata +--------+ | 397 +-------------------------------------------------------------+ 399 Figure 5: Split IPV4 forwarding service with Inter-FE LFB 401 As can be observed in Figure 5, the same details passed between IPV4 402 unicast LPM LFB and the IPV4 NH LFB are passed to the egress side of 403 the Inter-FE LFB. In addition an index for the inter-FE LFB 404 (interFEid) is passed as metadata. 406 The egress of the inter-FE LFB uses the received Inter-FE index 407 (InterFEid metadata) to select details for encapsulation when sending 408 messages towards the selected neighboring FE. These details will 409 include what to communicate as the source and destination FEID; in 410 addition the original metadata and any exception IDs may be passed 411 along with the original IPV4 packet. 413 On the ingress side of the inter-FE LFB the received packet and its 414 associated details are used to decide the packet graph continuation. 415 This includes what of the of the original metadata and exception IDs 416 to restore and what next LFB class instance to continue processing 417 on. In the illustrated case above, an IPV4 Nexthop LFB is selected 418 and metadata is passed on to it. 420 The ingress side of the inter-FE LFB consumes some of the information 421 passed (eg the destination FEID) and passes on the IPV4 packet 422 alongside with the ingress + NHinfo metadata to the IPV4 NextHop LFB 423 as was done earlier in both Figure 1 and Figure 2. 425 5. Generic Inter-FE connectivity 427 In this section we describe the generic encapsulation format in 428 Figure 6 as extended from the ForCES redirect packet format. We 429 intend for the described encapsulation to be a generic guideline of 430 the different needed fields to be made available by any used 431 transport for inter-FE LFB connectivity. We expect that for any 432 transport mechanism used, a description of how the different fields 433 will be encapsulated to be correlated to the information described in 434 Figure 6. The goal of this document is to provide ethernet 435 encapsulation, and to that end in Section 5.1 we illustrate how we 436 use the guidelines provided in this section to describe the fit for 437 inter-FE LFB interfacing over ethernet. 439 +-- Main ForCES header 440 | | 441 | +---- msg type = REDIRECT 442 | +---- Destination FEID 443 | +---- Source FEID 444 | +---- NEID (first word of Correlator) 445 | 446 +-- T = ExceptionID-TLV 447 | | 448 | +-- +-Exception Data ILV (I = exceptionID , L= length) 449 | | | | 450 | | | +----- V= Metadata value 451 | . | 452 | . | 453 | . +-Exception Data ILV 454 . 455 | 456 +- T = METADATA-TLV 457 | | 458 | +-- +-Meta Data ILV (I = metaid, L= length) 459 | | | | 460 | | | +----- V= Metadata value 461 | . | 462 | . | 463 | . +-Meta Data ILV 464 . 465 +- T = REDIRECTDATA-TLV 466 | 467 +-- Redirected packet Data 469 Figure 6: Packet format suggestion 471 o The ForCES main header as described in RFC5810 is used as a fixed 472 header to describe the Inter-FE encapsulation. 474 * The Source FEID field is mapped to the originating FE and the 475 destination FEID is mapped to the destination FEID. 477 * The first 32 bits of the correlator field are used to carry the 478 NEID. The 32-bit NEID defaults to 0. 480 o The ExceptionID TLV carries one or more exception IDs within ILVs. 481 The I in the ILV carries a globally defined exceptionID as per- 482 ForCES specification defined by IANA. This TLV is new to ForCES 483 and sits in the global ForCES TLV namespace. 485 o The METADATA and REDIRECTDATA TLV encapsulations are taken 486 directly from [RFC5810] section 7.9. 488 It is expected that a variety of transport encapsulations would be 489 applicable to carry the format described in Figure 6. In such a 490 case, a description of a mapping to interpret the inter-FE details 491 and translate into proprietary or legacy formatting would need to be 492 defined. For any mapping towards these definitions a different 493 document to describe the mapping, one per transport, is expected to 494 be defined. 496 5.1. Inter-FE Ethernet Connectivity 498 In this document, we describe a format that is to be used over 499 Ethernet. An existing implementation of this specification on top of 500 Linux Traffic Control [linux-tc] is described in [tc-ife]. 502 The following describes the mapping from Figure 6 to ethernet wire 503 encapsulation illustrated in Figure 7. 505 o When an NE tag is needed, a VLAN tag will be used. Note: that the 506 NEID as per Figure 6 is described as being 32 bits while a vlan 507 tag is 12 bits. It is however thought to be sufficient to use 12 508 bits within the scope of a LAN NE cluster. 510 o An ethernet type will be used to imply that a wire format is 511 carrying an inter-FE LFB packet. The ethernet type to be used is 512 0xFEFE (XXX: Note to editor, to be updated when issued by IEEE 513 Standards Association). 515 o The destination FEID will be mapped to the destination MAC address 516 of the target FEID. 518 o The source FEID will be mapped to the source MAC address of the 519 originating FEID. 521 o In this version of the specification, we only focus on data and 522 metadata. Therefore we are not going to describe how to carry the 523 ExceptionID information (future versions may). We are also not 524 going to use METADATA-TLV or REDIRECTDATA-TLV in order to save 525 shave off some overhead bytes. Figure 7 describes the payload. 527 0 1 2 3 528 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 529 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 530 | Outer Destination MAC Address (Destination FEID) | 531 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 532 | Outer Destination MAC Address | Outer Source MAC Address | 533 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 534 | Outer Source MAC Address (Source FEID) | 535 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 536 | Optional 802.1Q info (NEID) | Inter-FE ethertype | 537 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 538 | Metadata length | TLV encoded Metadata | 539 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 540 | TLV encoded Metadata ~~~..............~~ | 541 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 542 | Original Ethernet payload ~~................~~ | 543 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 545 Figure 7: Packet format suggestion 547 An outer Ethernet header is introduced to carry the information on 548 Destination FEID, Source FEID and optional NEID. 550 o The Outer Destination MAC Address carries the Destination FEID 551 identification. 553 o Outer Source MAC Address carries the Source FEID identification. 555 o When an NEID is needed, an optional 802.1Q is carried with 12-bit 556 VLANid representing the NEID. 558 o The ethernet type is used to identify the frame as inter-FE LFB 559 type. Ethertype 0xFEFE is to be used (XXX: Note, to editor update 560 when available). 562 o The 16-bit metadata length is used to described the total encoded 563 metadata length (including the 16 bits used to encode the metadata 564 length). 566 o One or more TLV encoded metadatum follows the metadata length 567 field. The TLV type identifies the Metadata id. ForCES IANA- 568 defined Metadata ids will be used. We recognize that using a 16 569 bit TLV restricts the metadata id to 16 bits instead of ForCES 570 define space of 32 bits. However, at the time of publication we 571 believe this is sufficient to carry all the info we need and 572 approach taken would save us 4 bytes per Metadatum transferred. 574 o The original ethernet payload is appended at the end of the 575 metadata as shown. 577 5.1.1. Inter-FE Ethernet Connectivity Issues 579 There are several issues that may arise due to using direct ethernet 580 encapsulation. 582 o Because we are adding data to existing ethernet frames, MTU issues 583 may arise. We recommend: 585 * To use large MTUs when possible (example with jumbo frames). 587 * Limit the amount of metadata that could be transmitted; our 588 definition allows for filtering of which metadata is to be 589 encapsulated in the frame. We recommend implementing this by 590 setting the egress port MTU to allow space for maximum size of 591 the metadata total size you wish to allow between FEs. In such 592 a setup, the port is configured to "lie" to the upper layers by 593 claiming to have a lower MTU than it is capable of. MTU 594 setting can be achieved by ForCES control of the port LFB(or 595 other config). In essence, the control plane making a decision 596 for the MTU settings of the egress port is implicitly deciding 597 how much metadata will be allowed. 599 o The frame may be dropped if there is congestion on the receiving 600 FE side. One approach to mitigate this issue is to make sure that 601 inter-FE LFB frames receive the highest priority treatment when 602 scheduled on the wire. Typically protocols that tunnel in the 603 middle box do not care and depend on the packet originator to 604 resend if the originator cares about reliability. We do not 605 expect to be any different. 607 o While we expect to use a unique IEEE-issued ethertype for the 608 inter-FE traffic, we use lessons learnt from VXLAN deployment xref 609 to be more flexible on the settings of the ethertype value used. 610 We make the ether type an LFB read-write component. Linux VXLAN 611 implementation uses UDP port 8472 because the deployment happened 612 much earlier than the point of RFC publication where the IANA 613 assigned udp port issued was 4789 [vxlan-udp]. For this reason we 614 make it possible to define at control time what ethertype to use 615 and default to the IEEE issued ethertype. We justify this by 616 assuming that a given ForCES NE is likely to be owned by a single 617 organization and that the organization's CE(or CE cluster) could 618 program all participating FEs via the inter-FE LFB (described in 619 this document) to recognize a private ethernet type used for 620 inter-LFB traffic (possibly those defined as available for private 621 use by the IEEE, namely: IDs 0x88B5 and 0x88B6) 623 6. Detailed Description of the Ethernet inter-FE LFB 625 The ethernet inter-FE LFB has two LFB input ports and three LFB 626 output ports. 628 +-----------------+ 629 Inter-FE LFB | | 630 Encapsulated | OUT2+--> decapsulated Packet + metadata 631 -------------->|IN2 | 632 Packet | | 633 | | 634 raw Packet + | OUT1+--> encapsulated Packet 635 -------------->|IN1 | 636 Metadata | | 637 | EXCEPTIONOUT +--> ExceptionID, packet + metadata 638 | | 639 +-----------------+ 641 Figure 8: Inter-FE LFB 643 6.1. Data Handling 645 The Inter-FE LFB can be positioned at the egress of a source FE. In 646 such a case an Inter-FE LFB instance receives via port IN1, raw 647 packet and metadata IDs from the preceding LFB instance. The 648 InterFEid metadatum MAY be present on the incoming raw data. The 649 processed encapsulated packet will go out on either LFB port OUT1 to 650 a downstream LFB or EXCEPTIONOUT port in the case of a failure. 652 The Inter-FE LFB can be positioned at the ingress of a receiving FE. 653 In such a case an Inter-FE LFB receives, via port IN2, an 654 encapsulated packet. Successful processing of the packet will result 655 in a raw packet with associated metadata IDs going downstream to an 656 LFB connected on OUT2. On failure the data is sent out EXCEPTIONOUT. 658 The Inter-FE LFB may use the InterFEid metadatum on egress of an FE 659 to lookup the IFETable table. The interFEid in such a case will be 660 generated by an upstream LFB instance (i.e one preceding the Inter-FE 661 LFB). The output result constitutes a matched table row which has 662 the InterFEinfo details i.e. the tuple {NEID,Destination FEID,Source 663 FEID, inter FE type, metafilters}. The metafilters lists define 664 which Metadatum are to be passed to the neighboring FE. 666 The component names used in describing processing are defined in 667 Section 6.2 669 6.1.1. Egress Processing 671 The egress Inter-FE LFB will receive an ethernet frame and 672 accompanying metadatum (including optionally the InterFEid metadatum) 673 at LFB port IN1. The ethernet frame may be 802.1Q tagged. 675 The InterFEid may be used to lookup IFETable table. If lookup is 676 successful, the inter-FE LFB will perform the following actions using 677 the resulting tuple: 679 o Increment statistics for packet and byte count observed. 681 o Walk each packet metadatum and apply against the relevant 682 MetaFilterList. If no legitimate metadata is found that needs to 683 be passed downstream then the processing stops and the packet is 684 allowed through as is. 686 o Check that the additional overhead of the outer header and 687 encapsulated metadata will not exceed MTU. If it does, increment 688 the error packet count statistics and return allowing the packet 689 to pass through. 691 o create the outer ethernet header which is a duplicate of the 692 incoming frame's ethernet header. The outer ethernet header may 693 have an optional 802.1q header (if one was included in the 694 original frame). 696 o If the NEID field is present (not 0) and the original header had a 697 vlan tag, replace the vlan tag on the outer header with the value 698 from the matched NEID field. If the NEID field is present (not 0) 699 and the original header did not have a vlan tag, create one that 700 matches the NEID field and appropriately add it to the outer 701 header. If the NEID field is absent or 0, do nothing. 703 o If the optional DSTFE is present, set the Destination MAC address 704 of the outer header with value found in the DSTFE field. When 705 absent, then the inner destination MAC address is used (at this 706 point already copied). 708 o If the optional SRCFE is present, set the Source MAC address of 709 the outer header with value found in the SRCFE field. If SRCFE is 710 absent then the inner source MAC address is used (at this point 711 already copied). 713 o If the optional IFETYPE is present, set the outer ethernet type to 714 the value found in IFETYPE. If IFETYPE is absent then the 715 standard ethernet type is used (XXX: Note to editor, to be 716 updated). 718 o encapsulate each allowed metadatum in a TLV. Use the Metaid as 719 the "type" field in the TLV header. The TLV should be aligned to 720 32 bits. This means you may need to add padding of zeroes to 721 ensure alignment. 723 o Update the Metadata length to the sum of each TLV's space + 2 724 bytes (for the Metadata length field 16 bit space). 726 The resulting packet is sent to the next LFB instance connected to 727 the OUT1 LFB-port; typically a port LFB. 729 In the case of a failed lookup or a zero-value InterFEid, (or absence 730 of InterFEid when needed by the implementation) the packet is sent 731 out unchanged via the OUT1 LFB Class instance port (typically towards 732 a Port LFB). 734 6.1.2. Ingress Processing 736 An inter-FE LFB packet is recognized by looking at the etherype 737 received on LFB instance port IN2. The IFETable table may be 738 optionally utilized to provide metadata filters. 740 o Increment statistics for packet and byte count observed. 742 o Look at the metadata length field and walk the packet data 743 extracting from the TLVs the metadata values. For each metadatum 744 extracted, the metaid is compared against the relevant IFETable 745 row metafilter list. If the metadatum is recognized, and is 746 allowed by the filter the corresponding implementation metadatum 747 field is set. If an unknown metadatum id is encountered, or if 748 the metaid is not found in the option allowed filter list the 749 implementation is expected to ignore it, increment the packet 750 error statistic and proceed processing other metadatum. 752 o Upon completion of processing all the metadata, the inter-FE LFB 753 instance resets the header to point to the original (inner) 754 ethernet header i.e skips the IFE header information. At this 755 point the the original ethernet frame that was passed to the 756 egress Inter-FE LFB at the source FE is reconstructed. This data 757 is then passed along with the reconstructed metadata downstream to 758 the next LFB instance in the graph. 760 In the case of processing failure of either ingress or egress 761 positioning of the LFB, the packet and metadata are sent out the 762 EXCEPTIONOUT LFB port with appropriate error id. Note that the 763 EXCEPTIONOUT LFB port is merely an abstraction and implementation may 764 in fact drop packets as described above. 766 6.2. Components 768 There are two LFB component populated by the CE. 770 The CE optionally programs LFB instances in a service graph that 771 require inter-FE connectivity with InterFEid values to correspond to 772 the inter-FE LFB IFETable table entries to use. 774 The first component is an array known as the IFETable table. The 775 array rows are made up of IFEInfo structure. The IFEInfo structure 776 constitutes: optional NEID, optional IFETYPE, optional Destination 777 FEID(DSTFE), optional Source FEID (SRCFE), optional array of allowed 778 Metaids (MetaFilterList). The table is looked up by a 32 bit index 779 passed from an upstream LFB class instance in the form of InterFEid 780 metadatum. 782 The second component(ID 2) is IFEStats table which carries the basic 783 stats structure bstats. The table index value used to lookup this 784 table is the same one as in IFETable table; in other words for a 785 table row index 10 in the IFETable table, its corresponding stats 786 will be found in row index of the IFEStats table. 788 6.3. Inter-FE LFB XML Model 790 793 795 796 EthernetAny 797 Packet with any Ethernet type 798 799 800 InterFEFrame 801 802 Packet with an encapsulate IFE Ethernet type 803 804 806 808 810 811 bstats 812 Basic stats 814 815 816 bytes 817 The total number of bytes seen 818 uint64 819 821 822 packets 823 The total number of packets seen 824 uint32 825 827 828 errors 829 The total number of packets with errors 830 uint32 831 832 834 836 837 IFEInfo 838 Describing IFE table row Information 839 840 841 NEID 842 843 The VLAN Id 12 bits part of the 802.1q TCI field. 844 845 846 uint16 847 848 849 IFETYPE 850 851 the ethernet type to be used for outgoing IFE frame 852 853 854 uint16 855 856 857 DSTFE 858 859 the destination MAC address of destination FE 860 861 862 byte[6] 863 864 865 SRCFE 866 867 the source MAC address used for the source FE 868 869 870 byte[6] 871 872 873 MetaFilterList 874 875 the allowed metadata filter table 876 877 878 879 uint32 880 881 883 884 886 888 889 890 InterFEid 891 892 Metadata identifying the index of the NexFE table 893 894 16 895 uint32 896 897 899 900 901 IFE 902 903 This LFB describes IFE connectivity parameterization 904 905 1.0 907 908 909 IN1 910 911 The input port of the egress side. 912 It expects any type of Ethernet frame. 913 914 915 916 EthernetAny 917 918 919 920 921 IN2 922 923 The input port of the ingress side. 924 It expects an inter-FE encapsulated Ethernet frame 925 with associated metadata. 926 927 928 929 InterFEFrame 930 931 932 InterFEid 933 934 935 937 939 941 942 OUT1 943 944 The output port of the egress side. 945 946 947 948 InterFEFrame 949 950 951 InterFEid 952 953 954 955 956 OUT2 957 958 The output port of the Ingress side. 959 960 961 962 EthernetAny 963 964 965 InterFEid 966 967 968 970 971 EXCEPTIONOUT 972 973 The exception handling path 974 975 976 977 EthernetAny 978 979 980 ExceptionID 981 InterFEid 982 983 984 986 988 990 991 IFETable 992 993 the table of all InterFE relations 994 995 996 IFEInfo 997 998 999 1000 IFEStats 1001 1002 the stats corresponding to the IFETable table 1004 1005 bstats 1006 1008 1010 1011 1012 1014 Figure 9: Inter-FE LFB XML 1016 7. Acknowledgements 1018 The authors would like to thank Joel Halpern and Dave Hood for the 1019 stimulating discussions. Evangelos Haleplidis contributed to 1020 improving this document. 1022 8. IANA Considerations 1024 This memo includes two IANA requests within the registry 1025 https://www.iana.org/assignments/forces 1027 The first request is for the sub-registry "Logical Functional Block 1028 (LFB) Class Names and Class Identifiers" to request for the 1029 reservation of LFB class name IFE with LFB classid 6112 with version 1030 1.0. 1032 The second request is for the sub-registry "Metadata ID" to request 1033 for the InterFEid metadata the value 0x00000010. 1035 9. IEEE Assignment Considerations 1037 This memo includes a request for a new ethernet protocol type as 1038 described in Section 5.1. 1040 10. Security Considerations 1042 This document does not alter either the ForCES model the ForCES Model 1043 [RFC5812] or the ForCES Protocol [RFC5810] As such, it has no impact 1044 on their security considerations. This document simply defines the 1045 operational parameters and capabilities of an LFB that performs LFB 1046 class instance extensions across nodes under a single administrative 1047 control. this document does not attempt to analyze the presence or 1048 possibility of security interactions created by allowing LFB graph 1049 extension on packets. Any such issues, if they exist, are for the 1050 designers of the particular data path, not the general mechanism. 1052 11. References 1054 11.1. Normative References 1056 [RFC3746] Yang, L., Dantu, R., Anderson, T., and R. Gopal, 1057 "Forwarding and Control Element Separation (ForCES) 1058 Framework", RFC 3746, April 2004. 1060 [RFC5810] Doria, A., Hadi Salim, J., Haas, R., Khosravi, H., Wang, 1061 W., Dong, L., Gopal, R., and J. Halpern, "Forwarding and 1062 Control Element Separation (ForCES) Protocol 1063 Specification", RFC 5810, March 2010. 1065 [RFC5811] Hadi Salim, J. and K. Ogawa, "SCTP-Based Transport Mapping 1066 Layer (TML) for the Forwarding and Control Element 1067 Separation (ForCES) Protocol", RFC 5811, March 2010. 1069 [RFC5812] Halpern, J. and J. Hadi Salim, "Forwarding and Control 1070 Element Separation (ForCES) Forwarding Element Model", 1071 RFC 5812, March 2010. 1073 11.2. Informative References 1075 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1076 Requirement Levels", BCP 14, RFC 2119, March 1997. 1078 [RFC6956] Wang, W., Haleplidis, E., Ogawa, K., Li, C., and J. 1079 Halpern, "Forwarding and Control Element Separation 1080 (ForCES) Logical Function Block (LFB) Library", RFC 6956, 1081 June 2013. 1083 [brcm-higig] 1084 "Higig", . 1086 [linux-tc] 1087 Hadi Salim, J., "Linux Traffic Control Classifier-Action 1088 Subsystem Architecture", netdev 01, Feb 2015. 1090 [tc-ife] Hadi Salim, J. and D. Joachimpillai, "Distributing Linux 1091 Traffic Control Classifier-Action Subsystem", netdev 01, 1092 Feb 2015. 1094 [vxlan-udp] 1095 "iproute2 and kernel code (drivers/net/vxlan.c)", 1096 . 1098 Authors' Addresses 1100 Damascane M. Joachimpillai 1101 Verizon 1102 60 Sylvan Rd 1103 Waltham, Mass. 02451 1104 USA 1106 Email: damascene.joachimpillai@verizon.com 1108 Jamal Hadi Salim 1109 Mojatatu Networks 1110 Suite 400, 303 Moodie Dr. 1111 Ottawa, Ontario K2H 9R4 1112 Canada 1114 Email: hadi@mojatatu.com