idnits 2.17.1 draft-qiang-netslices-gap-analysis-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 : ---------------------------------------------------------------------------- ** There are 57 instances of too long lines in the document, the longest one being 2 characters in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document doesn't use any RFC 2119 keywords, yet seems to have RFC 2119 boilerplate text. -- The document date (July 3, 2017) is 2460 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC5440' is defined on line 1091, but no explicit reference was found in the text == Outdated reference: A later version (-22) exists of draft-boucadair-connectivity-provisioning-protocol-14 == Outdated reference: A later version (-15) exists of draft-ietf-spring-segment-routing-12 == Outdated reference: A later version (-04) exists of draft-ooamdt-rtgwg-ooam-header-03 -- Obsolete informational reference (is this intentional?): RFC 2679 (Obsoleted by RFC 7679) -- Obsolete informational reference (is this intentional?): RFC 2680 (Obsoleted by RFC 7680) Summary: 1 error (**), 0 flaws (~~), 6 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 none L. Qiang, Ed. 3 Internet-Draft Huawei 4 Intended status: Informational P. Martinez-Julia 5 Expires: January 4, 2018 NICT 6 L. Geng 7 China Mobile 8 J. Dong 9 K. Makhijani 10 Huawei 11 A. Galis 12 University College London 13 S. Hares 14 Hickory Hill Consulting 15 S. Kuklinski 16 Orange 17 July 3, 2017 19 Gap Analysis for Transport Network Slicing 20 draft-qiang-netslices-gap-analysis-01 22 Abstract 24 This document presents network slicing differentiation from the non- 25 partition network or from simply partition of connectivity resources. 26 It lists 7 standardization gaps related to 4 key requirements for 27 network slicing in transport network. It also presents an analysis 28 of existing related work and other potential solutions on network 29 slicing. 31 This gap analysis document aims to provide a basis for future works 32 in transport network slicing. 34 Status of This Memo 36 This Internet-Draft is submitted in full conformance with the 37 provisions of BCP 78 and BCP 79. 39 Internet-Drafts are working documents of the Internet Engineering 40 Task Force (IETF). Note that other groups may also distribute 41 working documents as Internet-Drafts. The list of current Internet- 42 Drafts is at http://datatracker.ietf.org/drafts/current/. 44 Internet-Drafts are draft documents valid for a maximum of six months 45 and may be updated, replaced, or obsoleted by other documents at any 46 time. It is inappropriate to use Internet-Drafts as reference 47 material or to cite them other than as "work in progress." 48 This Internet-Draft will expire on January 4, 2018. 50 Copyright Notice 52 Copyright (c) 2017 IETF Trust and the persons identified as the 53 document authors. All rights reserved. 55 This document is subject to BCP 78 and the IETF Trust's Legal 56 Provisions Relating to IETF Documents 57 (http://trustee.ietf.org/license-info) in effect on the date of 58 publication of this document. Please review these documents 59 carefully, as they describe your rights and restrictions with respect 60 to this document. Code Components extracted from this document must 61 include Simplified BSD License text as described in Section 4.e of 62 the Trust Legal Provisions and are provided without warranty as 63 described in the Simplified BSD License. 65 Table of Contents 67 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 68 2. Terminology and Abbreviation . . . . . . . . . . . . . . . . 4 69 3. Overall Requirements in Network Slicing . . . . . . . . . . . 4 70 4. Network Slicing Specification . . . . . . . . . . . . . . . . 7 71 4.1. Description . . . . . . . . . . . . . . . . . . . . . . . 7 72 4.2. Related Work in IETF . . . . . . . . . . . . . . . . . . 8 73 4.2.1. YANG Data Models . . . . . . . . . . . . . . . . . . 8 74 4.2.2. Building NSS from Protocol Independent Traffic 75 Engineering Models . . . . . . . . . . . . . . . . . 9 76 5. Network Slicing Cross-Domain Coordination . . . . . . . . . . 10 77 5.1. Description . . . . . . . . . . . . . . . . . . . . . . . 10 78 5.2. Related Work in IETF . . . . . . . . . . . . . . . . . . 11 79 5.2.1. Autonomic Networking Integrated Model and Approach 80 (ANIMA) . . . . . . . . . . . . . . . . . . . . . . . 11 81 5.2.2. Connectivity Provisioning Negotiation Protocol (CPNP) 12 82 5.2.3. Abstraction and Control of Traffic Engineered 83 Networks (ACTN) . . . . . . . . . . . . . . . . . . . 13 84 5.3. Other Potential Solutions . . . . . . . . . . . . . . . . 14 85 6. Network Slicing Performance Guarantee and Isolation . . . . . 14 86 6.1. Description . . . . . . . . . . . . . . . . . . . . . . . 14 87 6.2. Related Work in IETF . . . . . . . . . . . . . . . . . . 15 88 6.2.1. Virtual Private Networks . . . . . . . . . . . . . . 15 89 6.2.2. NVO3 . . . . . . . . . . . . . . . . . . . . . . . . 15 90 6.2.3. RSVP-TE . . . . . . . . . . . . . . . . . . . . . . . 15 91 6.2.4. Segment Routing . . . . . . . . . . . . . . . . . . . 16 92 6.2.5. Deterministic Networking . . . . . . . . . . . . . . 16 93 6.2.6. Flexible Ethernet . . . . . . . . . . . . . . . . . . 17 94 7. Network Slicing OAM with Customized Granularity . . . . . . . 17 95 7.1. Description . . . . . . . . . . . . . . . . . . . . . . . 17 96 7.2. Related Work in IETF . . . . . . . . . . . . . . . . . . 18 97 7.2.1. Overview of OAM tools . . . . . . . . . . . . . . . . 18 98 7.2.2. Overlay OAM . . . . . . . . . . . . . . . . . . . . . 19 99 7.2.3. Service Function Chaining . . . . . . . . . . . . . . 19 100 7.2.4. Slice Identification . . . . . . . . . . . . . . . . 19 101 8. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 102 9. Security Considerations . . . . . . . . . . . . . . . . . . . 21 103 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 21 104 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 22 105 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 22 106 12.1. Normative References . . . . . . . . . . . . . . . . . . 22 107 12.2. Informative References . . . . . . . . . . . . . . . . . 22 108 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 26 110 1. Introduction 112 Network slicing is an approach to enable flexible isolation of 113 network resources and functions for dedicated services, providing a 114 certain level of customization and quality guarantee. It establishes 115 customized dedicated network upon a common infrastructure for 116 vertical industries with flexible design of functions, different 117 performance requirements, system isolation and OAM tools. 119 Several SDOs have investigated network slicing. To list a few: NGMN 120 initiated a study of network slicing in the context of 5G from the 121 mobile network point of view [NGMN-2016]. Around the same time ITU-T 122 IMT 2020 and ITU-T SG13 studied network softwarization that also 123 included network slicing concept. ITU-T has issued a number of 124 recommendations, such as: Gap Analysis [IMT2020-2015], Network 125 Softwarization [IMT2020-2016], Terms & Definitions [IMT2020-2016bis]. 126 Open Network Foundation (ONF) has developed a recommendation on 127 applying SDN architecture to Network Slicing [ONF-2016]. Finally, 128 3GPP standards development for 5G includes network slicing in radio 129 access and core networks. 3GPP issued TS 23.501 [TS23-501] about the 130 system architecture for 5G in 2017. BBF started the project SD-406 131 focusing on the end-to-end architecture enhancement and requirements 132 gathering for transport networks. Although these SDOs have done a 133 lot of work, potential requirements especially in the transport 134 network and end-to-end enabling need to be investigated in order to 135 elicit and identify the technical gaps in IETF for transport network 136 slicing. 138 In order to establish a network slice that meets various customer's 139 demands, an infrastructure owner needs to understand how these 140 demands map with the available network resources and accessible 141 capabilities. This also requires end-to-end coverage and inter- 142 domain coordination. Meanwhile, the slice provider provides 143 customized OAM to the tenants under provisioning. Slicing OAM 144 approach is a fundamental capability to guarantee stable, effective 145 and reliable services for the vertical industries. It is also 146 expected to be capable of operations with customized granularity 147 levels that provides robust management flexibilities. 149 This document presents the identified key requirements and 150 investigates potential technical gaps accordingly. To assist 151 understanding of this document, Section 2 outlines the terminology. 152 Section 3 introduces overall requirements of network slicing. 153 Sections 4~7 illustrates resource specification, end-to-end 154 consideration, performance guarantee and OAM concerns respectively. 155 Section 8 summarizes the identified gaps. 157 2. Terminology and Abbreviation 159 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 160 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 161 document are to be interpreted as described in RFC 2119. 163 All of the network slicing related words used in this document are to 164 interpreted as described in [NS-Framework] 166 3. Overall Requirements in Network Slicing 168 This section introduces 4 key requirements of network slicing derived 169 from [NS-UseCase] as shown in Table 1. These 4 requirements are 170 organized according to a general network slice working process as 171 shown in Figure 1: 173 1: describe network slicing resource/functions and capture 174 requirements (Req. 1) 176 2: network slicing cross-domain coordination (Req. 2) 178 3: construct a performance guaranteed and isolated end-to-end 179 network slice (Req. 3) 181 4: provide necessary Operation & Maintenance & Administration (OAM) 182 (Req. 4) 184 +----------------------------------------------------------+ 185 | network slice management and orchestration <-----+ 186 +----------------------^-------^---------------------------+ | 187 | | resource/functions 188 | OAM | specification 189 | | | 190 +------------------------v-------+------------------------------+ | 191 | abstracted network slice instance 1 | | 192 +--------------------------------+------------------------------+ | 193 | | 194 +--------------------------------v------------------------------+ | 195 | abstracted network slice instance 2 | | 196 +---------------------------------------------------------------+ | 197 | 198 | 199 +---------+ +---------+ +---------+ | 200 |NS-Domain| cross-domain |NS-Domain| cross-domain |NS-Domain<-----+ 201 | Manager <--------------> Manager <--------------> Manager | 202 +---------+ coordination +---------+ coordination +---------+ 204 +---------+ +---------+ +---------+ 205 | | | | | | 206 +-+---------+--------------+---------+--------------+---------+-+ 207 | network slice instance 1 <---+ 208 +-+---------+--------------+---------+--------------+---------+-+ | 209 | Domain 1| | Domain 2| | Domain 3| isolation 210 +-+---------+--------------+---------+--------------+---------+-+ | 211 | network slice instance 2 <---+ 212 +-+---------+--------------+---------+--------------+---------+-+ 213 | | | | | | 214 +---------+ +---------+ +---------+ 216 Figure 1: Illustration of Key Requirements 218 Table 1: Requirement Association 220 +-----------------------------------------+------------------------------+ 221 | Requirements Illustrated in NS UseCase | Extracted KEY Requirements | 222 +-----------------------------------------+------------------------------+ 223 |1) Resource Reservation | | 224 |2) Abstraction | Req 1. Network Slicing | 225 |3) Multi-Access Knowledge | | 226 |4) Multi-Dimensional Service Vertical | Specification | 227 |5) Agile Resource Adjustment | | 228 +-----------------------------------------+------------------------------+ 229 |6) Multi-Domain Coordination | Req 2. Network Slicing | 230 | | Cross-Domain | 231 |7) Resource Assurance | Coordination | 232 +-----------------------------------------+------------------------------+ 233 | | Req 3. Network Slicing | 234 |8) Performance/Operation Isolation | Performance Guarantee | 235 | | and Isolation | 236 +-----------------------------------------+------------------------------+ 237 |9) Independent Slice Management Plane | Req 4. Network Slicing OAM | 238 | Reliability | | 239 +-----------------------------------------+------------------------------+ 241 Table 1: Requirements Association 243 o Req 1. Network Slicing Specification (NSS) - The management 244 systems of both network slice providers and operators need to know 245 what and how much resources/network functions they have, so that 246 they can accurately and abstractedly describe the available 247 resources/network functions to tenants or peers. The objective of 248 NSS is to deliver the network slicing requests without incurring 249 any over-utilization of resources. In order to cooperate and 250 provide consistent network slicing service, the way that 251 resources/network functions are described should be homogeneous 252 and compatible among all of the involved technology-specific 253 domains, provides, and slicing platforms. 255 o Req 2. Network Slicing Cross-Domain Coordination (NS-CDC) - From 256 terminal to server (or other terminal), an end-to-end network 257 slice will involve different infrastructural domains (e.g., AN, 258 TN, CN, etc. ) that may be owned by different providers/operators. 259 Each infrastructural domain may be further divided into different 260 administrative domains. That is an end-to-end slice is a logical 261 entity composed by multiple separated components, and the cross- 262 domain coordination is a way to integrate these components 263 together. 265 o Req 3. Network Slicing Performance Guarantee and Isolation (NS- 266 PGI) - In order to enable the safe, secure, privacy-preservation 267 service for multi-tenancy on a common physical network, the 268 isolation among network slices in each of the 269 Data/Control/Management/Service planes are needed. Furthermore, 270 network slices that provide differentiated services usually 271 require different resources. The resources allocated to a network 272 slice must be able to guarantee the service performance 273 requirement. 275 o Req 4. Network Slicing OAM (NS-OAM) - On one end of the spectrum 276 we have those operators that will require a finalized service that 277 they will simply commercialize. On the other end we have those 278 operators that may want to fine-tune all the low-level aspects of 279 the network resources that form their system or service. 280 Moreover, in the middle there is plenty of room for variations. 281 Therefore, the underlying network layers must offer different 282 levels of granularity for the management of their resources, that 283 the upper layer operators can choose according to their needs and 284 objectives. 286 4. Network Slicing Specification 288 4.1. Description 290 Network Slicing Specification (NSS) is meant to describe the network 291 slicing resources and capture requirements from tenants or peer 292 networks to characterize the service expected to be delivered by a 293 network. These requirements include (non-exhaustive): reachability 294 scope (e.g., limited scope, Internet-wide), direction, bandwidth 295 requirements, performance metrics (e.g., one-way delay [RFC2679], 296 loss [RFC2680], or one-way delay variation [RFC3393]), protection and 297 high-availability guidelines (e.g., uRLLC service restoration in less 298 than 50 ms, 100 ms, or 1 second), traffic isolation constraints, and 299 flow identification. NSS is used by a network provider to decide 300 whether existing network slice instances can be reused or (some of 301 them) even combined, or if another network slice instance is needed 302 for a given service. 304 Technology-specific actions are then derived from the technology- 305 agnostic requirements depicted in an NSS. Such actions include 306 configuration tasks and operational procedures. A standard 307 definition of NSS is needed to facilitate the dynamic/ automated 308 negotiation procedure of NSS parameters, but also to homogenize the 309 processing of service requirements. 311 To explain by an example, a network slice may cross multiple domains: 313 o A cloud deployed, NFV enabled, chain of network functions in a 314 virtualized 5G core. 316 o A segment routing [I-D.ietf-spring-segment-routing] based IGP 317 network transport/aggregation or slice-specific application 318 functions. 320 o A PCE [RFC4655] monitored TE-tunnel with ingress and egress 321 points. 323 o Optical, carrier Ethernet or cellular networks. 325 The network slice is a combination of the above technologies. It 326 creates a compelling need for a common resource specification 327 interface across these domains. 329 4.2. Related Work in IETF 331 4.2.1. YANG Data Models 333 As rightfully discussed in [I-D.wu-opsawg-service-model-explained], 334 the IETF has already published several YANG data models that are used 335 to model monolithic functions as well as very few services (e.g., 336 L2SM, L3SM, EVPN). These models may be used in the context of 337 network slicing if corresponding technologies are required for a 338 given network slice, but none of them can be used to model an NSRD. 340 [RFC7297] describes the Connectivity Provisioning Profile (CPP) and 341 proposes a CPP template to capture connectivity requirements to be 342 met within a service delivery context. Such a generic CPP template 343 is meant to 345 o facilitate the automation of the service negotiation and 346 activation procedures, thus accelerating service provisioning; 348 o set (traffic) objectives of Traffic Engineering (TE) functions and 349 service management functions; 351 o improve service and network management systems with 'decision- 352 making' capabilities based upon negotiated/offered CPPs. 354 [RFC7297] may be considered as a candidate specification for NSRD. 355 Releasing a RFC7297-bis to take into account specific requirements 356 from network slicing is needed. Since [RFC7297] may not be 357 implemented by all providers, the [SLA-Exchange] may be adopted to 358 implement indirect SLA negotiation and SLA events report. 359 [SLA-Exchange] provides an in-band method to exchange the SLA 360 parameters, and then by the receiving devices to translate SLA in 361 technical specific provisioning languages. However, there still does 362 not exist any standard protocol to translate SLA agreements into 363 technical clauses and configurations. 365 4.2.2. Building NSS from Protocol Independent Traffic Engineering 366 Models 368 The NSS requirement for reachability, direction, bandwidth 369 requirements, performance metrics, traffic isolation constraints, and 370 flow identification can be built utilizing protocol which can perform 371 operations (read, write, notification, actions (aka rpcs)) on a yang 372 service layer that supports these traffic engineer and resource 373 definition at the service layers. The network slicing service data 374 model can extend existing work in the TEAS and I2RS working group for 375 protocol-independent topology models. These models support 376 configuration or the dynamic datastores defined in [NMDA] which will 377 be abbreviated as NMDA in this section. This section provides the 378 detail on how the NSS can be built from these models and the RESTCONF 379 protocol. 381 4.2.2.1. Basic Topology Model 383 The basic topology model is defined in [I2RS-Yang] to include a 384 service layer. This topology model is protocol independent and can 385 be utilized as a configuration data model or a dynamic datastores 386 model. The configuration data model must abide by the configuration 387 persistence and referential requirements. The dynamic datastores do 388 not need to abide by the same requirements as the configuration 389 datastore. I2RS is defining a dynamic datastores reference model for 390 a data store which ephemeral. The network slices may want to use 391 configuration, ephemeral datastores, or define a third type of 392 dynamic datastores. The I2RS WG provides a place to collaborate this 393 work on the dynamic datastores. 395 4.2.2.2. TEAS Model Utilization of Basic Topology Model 397 The TEAS topology model [TE-Yang] provides a general description of a 398 Traffic engineering model that provides: 400 o abstract topologies with TE constraints (bandwidth, delay metrics, 401 links to lower layers, some traffic isolation constraints, and 402 some link identifiers); 404 o templates for links or resources; 406 o functionality to read, write, notification, and rpcs. 408 Options that need to be consider are: 410 Augmenting TEAS - The TEAS models provide substantial traffic 411 engineering. It was envisioned in the early topology model that a 412 service resource model would be part of the service layer. This 413 work was delayed until the maturation of the service requirements 414 from L2VPN, L3VPN, and EVPN plus the maturation of resource 415 requirements from 5G. Network slicing provides a good application 416 use case for this work. 418 Why not Augment TEAS - The TEAS models are TE specific, lack of 419 the abstraction for Layer 3+ resources. 421 Dynamic models to combine TEAS models for network-slicing - The 422 network slicing controller operating across domains may wish to 423 create a multiple-domain data model based on the service layer 424 data models exposed by different providers. These service models 425 would not need to be configured, but only learned as providers 426 exchange data with one another. The rules for combining these 427 models could be defined as part of the dynamic datastore for 428 network-slicing. 430 Protocol within a domain - The RESTCONF and NETCONF protocol can 431 support read, write, notification and actions (rpcs) within a 432 domain. 434 Protocol across domains: The RESTCONF protocol currently supports 435 Configuration protocols and 90% of the dynamic datastores. The 436 RESTCONF protocol is being enhanced to support the push of 437 telemetry messages. The RESTCONF protocol could be used to 438 exchange a specific Yang network-slicing service-layer topology 439 (TE and Resources) and for the I2NSF security capabilities between 440 domains. 442 If a multicast of telemetry data is required between domains, then 443 the push model for telemetry information or the IPFIX protocol may 444 be utilized. 446 5. Network Slicing Cross-Domain Coordination 448 5.1. Description 450 The network slicing cross-domain coordination (NS-CDC) requirement 451 includes the following aspects: 453 o Network slice resource/functions coordination: for example, a 454 tenant requests for a network slice with at most 10 ms latency 455 from terminal to server. Different infrastructure/administrative 456 domains should coordinate and negotiate to reach an agreement such 457 as RAN provides at most 2 ms service, TN domain I provides at most 458 4ms service, TN domain II provides at most 2 ms service and CN 459 provides at most 2 ms service; 461 o Configuration information coordination: for example, for a given 462 TN domain, the configuration information such as VLAN ID, remote 463 IP address, physical port ID, etc. need to be coordinated with 464 other TN domains; 466 o Other coordination: for example, RAN (or other access network) 467 needs to notify TN about the information of new attachment point 468 when user moves. 470 From terminal to server, an end-to-end network slice will involve 471 different infrastructure domains (e.g., RAN, TN and CN). An 472 infrastructure domain may be further divided into multiple domains 473 due to geographic isolation, administrative isolation and other 474 reasons. There are two ways to enable an end-to-end network slice: 475 based on a common platform or based on cross-domain coordination. 477 If all of the involved domains belong to the same operator or the 478 same operator union, the common platform solution may be work. In 479 this case, all of the domain controllers only need to communicate 480 with the common platform, and follow the coordination management of 481 this common platform. Whilst the most common case is that the 482 domains belong to different owners/operators/administrators, making 483 it difficult to realize such a common platform. Consequently, the 484 cross-domain coordination will be essential throughout the whole 485 lifecycle of an end-to-end network slice. 487 5.2. Related Work in IETF 489 There are some related works studies the inter-operation/coordination 490 between different entities. Coordination of different components of 491 a slice requires automation. It can be achieved either by 493 1. Coordination protocols such as ANIMA, CPNP 495 2. Or through abstraction and corresponding interfaces as in ACTN. 497 This subsection will briefly review these related work to provide a 498 basis for the gap analysis. 500 5.2.1. Autonomic Networking Integrated Model and Approach (ANIMA) 502 Autonomic Networking Integrated Model and Approach (ANIMA) WG 503 provides a series of tools for distributed and automatic management, 504 which includes: Generic Autonomic Signaling Protocol (GRASP), 505 Autonomic Networking Infrastructure (ANI), etc. 507 GRASP [ANIMA-GRASP] is a protocol for the negotiation between ASAs 508 (Autonomic Service Agent). In GRASP, ASAs could be considered as 509 "APPs" installed on a device. Different ASAs fulfill different 510 management tasks such as parameter configuration, service delivery, 511 etc. Based on GRASP, the same purpose ASAs that installed on 512 different devices are able to inter-operate and negotiate with each 513 other. Network slicing could make use of GRASP for the coordination 514 among devices in the underlying infrastructure layer, as well as the 515 negotiation among different domain managers. However, the security 516 issue incurred by cross-domain usage should be fixed in GRASP. 518 ANI [ANI] is a technical packet consisting of BootStrap (for 519 authentication, domain certification distribution, etc.), ACP (a 520 separate control plane), and GRASP (for control message 521 coordination). ANI could be used to construct the management tunnel 522 among devices in underlying infrastructure layer within a single 523 domain. While the network slicing and cross-domain oriented 524 extensions are necessary. 526 5.2.2. Connectivity Provisioning Negotiation Protocol (CPNP) 528 [I-D.boucadair-connectivity-provisioning-protocol] defines the 529 Connectivity Provisioning Negotiation Protocol (CPNP) that is meant 530 to dynamically exchange and negotiate connectivity provisioning 531 parameters, and other service-specific parameters, between a Customer 532 and a Provider. CPNP is a tool that introduces automation in service 533 negotiation and activation procedures, thus fostering the overall 534 service provisioning process. 536 CPNP runs between a Customer and a Provider carrying service orders 537 from the Customer and respective responses from the Provider to the 538 end of reaching a connectivity service provisioning agreement. As 539 the services offered by the Provider are well-described, by means of 540 the CPP template, the negotiation process is essentially a value- 541 settlement process, where an agreement is pursued on the values of 542 the commonly understood information items (service parameters) 543 included in the service description template. 545 The protocol is transparent to the content that it carries and to the 546 negotiation logic, at Customer and Provider sides, that manipulates 547 the content. 549 The protocol aims at facilitating the execution of the negotiation 550 logic by providing the required generic communication primitives. 552 CPNP can be used in the context of network slicing to request for 553 network resources together with a set of requirements that need to be 554 satisfied by the Provider. Such requirements are not restricted to 555 basic IP forwarding capabilities, but may also include a 556 characterization of a set of service functions that may be invoked. 558 5.2.3. Abstraction and Control of Traffic Engineered Networks (ACTN) 560 ACTN [TEAS-ACTN] is an information model proposed by TEAS WG, which 561 enables the multi-domain coordination in Traffic Engineering (TE) 562 network. In order to enable network slicing in transport networks, 563 portion of transport domain will need to be engineered. In 564 particular, building a TE entity and stitching service for this 565 entity is within the scope of ACTN. As an end-to-end network slicing 566 solution, ACTN is able to provide cross-domain coordination. In 567 ACTN, each physical transport network domain is under the control of 568 a Physical Network Controller (PNC) as shown in Figure 2. A Multi- 569 Domain Service Coordinator (MDSC) controls multiple PNCs. Although 570 the MDSCs may form a hierarchical structure, a hierarchical MDSC can 571 still be regarded as a logical common platform. As Section 5.1 572 discussed, such a common platform solution has a strict presumption 573 that all domains are assumed to follow a common coordination 574 management. 576 While ACTN does carry out network slicing-related work, some proposed 577 concepts are similar the concepts of today's network slicing: in 578 particular, the virtual network (VN) is similar to a slice instance. 579 ACTN enables VN based on LSP technique, different LSP tunnels 580 correspond to different VNs. However, ACTN focuses on resource 581 abstraction and management on Layer 2 and Layer 1. For transport 582 network slicing, resources abstraction and management on Layer 3+ 583 (e.g., IP routing table, etc.) may also be necessary but have not 584 been addressed by ACTN. 586 +-------+ +-------+ +-------+ 587 | CNC-A | | CNC-B | | CNC-C | 588 +---+---+ +---+---+ +---+---+ 589 | | | 590 +-------\ |CMI /------+ 591 \ | / 592 +---------+----------+ 593 | (Hierarchical)MDSC | 594 +---------+----------+ 595 / | \ 596 +-------+ |MPI +---------+ 597 | | | 598 +---+---+ +-------+ +----+--+ 599 | PNC | | PNC | | PNC | 600 +-------+ +-------+ +-------+ 602 Figure 2: A Three-tier ACTN Control Hierarchy 604 5.3. Other Potential Solutions 606 5G Exchange (5GEx) [FGEx] is a 5G-PPP project which aims to enable 607 cross-domain orchestration of services over multiple administrations 608 or over multi-domain single administration networks. The main 609 infrastructure considered in 5GEx is the NFV/SDN compatible software 610 defined infrastructure, which limits the scope of network slicing to 611 SDN based architecture. 613 6. Network Slicing Performance Guarantee and Isolation 615 6.1. Description 617 Network slicing is expected to enable the deployment of various 618 services with diverse requirements, independently on a common 619 physical network. Each network slice is characterized by particular 620 service requirements, which usually are expressed using in the form 621 of several key performance indicators (KPIs) such as bandwidth, 622 latency, jitter, packet loss, etc., and different degrees of 623 isolation. Isolation requirements include performance isolation, 624 which means performance guarantee are maintained regardless of 625 activity in other slices, as well as secure isolation (e.g., 626 including privacy), and management (or OAM) isolation. Additionally, 627 performance isolation in network slicing has to maintain while 628 scaling up or down computing capabilities of a slice (i.e., for 629 elastic scaling). Moreover, since IoT is also a use case for NS, and 630 since some IoT applications are sensitive to data plane or bits on 631 wire overheads, data path encapsulation in the form of labels, VLANs, 632 VxLANs should be optional, or minimized for those cases. 634 As we will discuss in the detailed sections below, each of these 635 technologies can address some but not all performance and isolation 636 requirements: 638 o RSVP-TE, Segment Routing (SR), DETNET, FlexE are mostly related to 639 performance guarantee and performance isolation requirements 641 o Virtual Private Networks (VPN), NVO3 are mostly related to 642 security and management isolation requirements 644 A Network Slicing solution, to support performance guarantee and 645 isolation requirements, will therefore need to merge in some way 646 characteristics from these two families of technologies, through the 647 combination of (possibly enhanced) existing technologies and/or 648 specifically developed ones. We can also consider the possibility 649 that multiple such technology stacks may be deployed in different 650 domains, and rely on cross-domain coordination, as described 651 inSection 5, to form a single abstracted network slice. 653 6.2. Related Work in IETF 655 6.2.1. Virtual Private Networks 657 VPN technologies such as L3VPN [RFC4364], L2VPN [RFC4664], EVPN 658 [RFC7432], etc. have been widely deployed to provide different 659 virtual networks on common service provider networks. Although VPNs 660 can provide logically separated routing/bridging domains between 661 different VPN customers, essentially it is an overlay network 662 technology with little control of the network resources, so it is 663 challenging for VPN to meet the performance and isolation requirement 664 of some emerging application scenarios such as industrial verticals. 665 VPNs essentially are private networks of enterprises by connecting 666 remote sites. The following two issues illustrate limitations of 667 VPNs for network slicing: 669 o An end-to-end VPN tunnel competes with other traffic in the 670 network and end-to-end network resource policies cannot be 671 guaranteed. 673 o The reachability and resource reservation protocols are not 674 tightly integrated and often solutions require centralized PCE-P 675 like methods. 677 6.2.2. NVO3 679 [NVO3-WG] defines several network encapsulations which support the 680 network virtualization and multi-tenancy in the data center networks. 681 Similar to the VPN technologies of service provider networks, NVO3 is 682 also an overlay network technology, which relies on the performance 683 characteristics provided by the IP-based underlay networks. Thus 684 NVO3 may not meet by itself the performance and isolation 685 requirements of network slicing. 687 6.2.3. RSVP-TE 689 RSVP-TE [RFC3209] is the signaling protocol to establish end-to-end 690 traffic-engineered Label Switched Paths (LSPs). It can reserve the 691 required link bandwidth along an end-to-end path for specific network 692 flows, which is suitable for services with particular requirement on 693 traffic bandwidth. RSVP-TE LSPs can be used as the underlay tunnels 694 of the VPN service connections. However, the requirement of some 695 emerging services is not only about traffic bandwidth, but also has 696 quite strict requirement on latency, jitter, etc. Such requirements 697 can hardly be met with existing RSVP-TE. 699 6.2.4. Segment Routing 701 [I-D.ietf-spring-segment-routing] provides the ability to specify a 702 traffic-engineered path by the source node of data packets. It can 703 provide traffic-engineering features comparable to RSVP-TE with 704 better scalability, by eliminating the per-path state in the transit 705 network nodes. It is therefore a candidate method of creating an 706 NSI, mapping a packet into an NSI and specifying the passage of the 707 packet through the resources dedicated to the NSI. Further study 708 will be required to determine if/how SR as designed today can be used 709 as a core technology for building an NSI. With respect to 710 performance guarantee and isolation, some further investigation may 711 be needed to understand whether SR can provide the same or better 712 performance characteristics as RSVP-TE. In addition, it is not clear 713 whether SR-based LSPs can provide the guaranteed latency and jitter 714 performance required by network slicing. 716 6.2.5. Deterministic Networking 718 [DETNET-WG] is working on the deterministic data paths over layer 2 719 and layer 3 network segments. Such deterministic paths can provide 720 identified flows with extremely low packet loss rates, low packet 721 delay variation (jitter) and assured maximum end-to-end delivery 722 latency. This is accomplished by dedicating network resources such 723 as link bandwidth and buffer space to DetNet flows and/or classes of 724 DetNet flows. DetNet also aims to provide high reliability by 725 replicating packets along multiple paths. It is a characteristic of 726 DetNet that it is concerned solely with worst-case values for the 727 end-to-end latency. 729 The primary target of DetNet is real-time systems and as such 730 average, mean, or typical latency values are not protected, because 731 they do not affect the ability of a real-time system to perform their 732 tasks. This contrasts with a normal priority-based queuing scheme 733 which will give better average latency to a data flow than DetNet, 734 but, on the other side, the worst-case latency can be essentially 735 unbounded. As such DetNet seems to be a useful technique that may be 736 applied to either a complete NSI, or to part of the traffic within an 737 NSI to address the emerging low latency requirement for real time 738 application. 740 DetNet can therefore address some of the requirements of NS. It was 741 however not designed with network slicing in mind, which means a 742 mapping between an NSI and a DetNet service may need to be defined. 744 6.2.6. Flexible Ethernet 746 [FLEXE-1.0] was initially defined by Optical Internetworking Forum 747 (OIF) as an interface technology which allows the complete decoupling 748 of the Media Access Control layer (MAC) data rates and the standard- 749 based Ethernet Physical layer (PHY) rates. The channelization 750 capability of FlexE can be used to partition a FlexE interface into 751 several independent sub-interfaces, which can be considered as a 752 useful component for the slicing of network interfaces. Currently 753 there is ongoing work in IETF to define the control plane framework 754 for FlexE [FlexE-FWK], which aims to identify the routing and 755 signaling extensions needed for establishing FlexE-based end-to-end 756 LSPs in IP/MPLS networks. 758 7. Network Slicing OAM with Customized Granularity 760 7.1. Description 762 In accordance with [RFC6291], OAM is used to denote the following: 764 o Operations: refer to activities that are undertaken to keep the 765 network and the services it deliver up and running. It includes 766 monitoring the underlying resources and identifying problems. 768 o Administration: refer to activities to keep track of resources 769 within the network and how they are used. 771 o Maintenance: refer to activities to facilitate repairs and 772 upgrades. Maintenance also involves corrective and preventive 773 measures to make the managed network run more effectively, e.g., 774 adjusting configuration and parameters. 776 As per [RFC6291], network slicing provisioning operations are not 777 considered as part of OAM. Provisioning operations are discussed in 778 other sections. 780 Maintaining automatically-provisioned slices within a network raises 781 the following requirements: 783 o Ability to run OAM activities on a provider's customized 784 granularity level. In other words, ability to run OAM activities 785 at any level of granularity that a service provider see fit. In 786 particular: 788 * Per slice OAM: An operator must be able to execute OAM tasks on 789 a per slice basis. 791 * Per domain OAM: These tasks can cover the "whole" slice within 792 a domain or a portion of that slice (for troubleshooting 793 purposes, for example). 795 * Per service OAM: When a given slice is shared among multiple 796 services/customers, an operator must be able to execute (per- 797 slice) OAM tasks for a particular service or customer. 799 * For example, OAM tasks can consist in tracing resources that 800 are bound to a given slice, tracing resources that are invoked 801 when forwarding a given flow bound to a given network slice, 802 assessing whether flow isolation characteristics are in 803 conformance with the NS Resource Specification, or assessing 804 the compliance of the allocated slice resource against flow/ 805 customer requirements. 807 * An operator must be able to enable differentiated failure 808 detect and repair features for a specific/subset of network 809 slices. For example, a given slice may require fast detect and 810 repair mechanisms (e.g., as a function of the nature of the 811 traffic (pattern) forwarded through the NS), while others may 812 not be engineered with such means. 814 o Ability to automatically discover the underlying service functions 815 and the slices they are involved in or they belong to. 817 o Ability to dynamically discover the set of network slicing that 818 are enabled within a network. Such dynamic discovery capability 819 facilitates the detection of any mismatch between the view 820 maintained by the control plane and the actual network 821 configuration. When mismatches are detected, corrective actions 822 must be undertaken accordingly. 824 o Ability to efficiently OAM on shared resources. If multiple 825 network slices share some resources, the same kind of OAM 826 operations from different network slices should be performed only 827 once for efficiency. For example, several network slices share a 828 link. We only need to execute once status query, and directly 829 return the queried result to other status query requests. 831 7.2. Related Work in IETF 833 7.2.1. Overview of OAM tools 835 The reader may refer to [RFC7276] for an overview about available OAM 836 tools. These technology-specific tools can be reused in the context 837 of network slicing. Providers that deploy network slicing 838 capabilities should be able to select whatever OAM technology- 839 specific feature that would be address their needs. No gap that 840 would legitimate specific requirements has been identified so far. 842 7.2.2. Overlay OAM 844 [I-D.ooamdt-rtgwg-ooam-header] specifies a generic OAM header that 845 can be used if overlay technologies are enabled. Obviously, this 846 effort can be reused in the context of network slicing when overlay 847 techniques are in use. Nevertheless, For slice designs that do not 848 assume an overlay technology, OAM packets must be able to fly over 849 the appropriate slice and for a given service/customer. This is 850 possible by reusing some existing tools if and only if no specific 851 fields are required (e.g., carry a slice identifier as Req. 5 852 stated). 854 7.2.3. Service Function Chaining 856 SFC WG [SFCWG] is chartered to describe data plane service 857 encapsulation, control and manageability aspects of service 858 functions. Extensions that will be specified by the SFC WG will be 859 reused in the context of network slicing. Nevertheless, The current 860 charter of the WG does not imply work on the automated discovery of 861 SF instances and their capabilities, nor the automatic discovery of 862 control elements. An additional specification effort is therefore 863 required in this area. 865 7.2.4. Slice Identification 867 A network slice data plane, may or may not follow traditional data 868 plane tagging/labeling. However, each network element (router/ 869 switch) still has to classify an incoming packet and associated with 870 the slice instance for proper treatment. Network slice instance 871 identification is essential for network element to make local 872 decisions on forwarding policies, QoS mechanism and etc. The 873 performance requirements of a network slice instance can therefore 874 been met by making the correct decision. Meanwhile, it is also 875 important for OAM so that configuration and provisioning can be 876 delicately performed to particular network slice instances by their 877 identifications. 879 For flow identification, many existing technologies provide mature 880 solutions. These approaches might be able to be re-used in network 881 slicing by adding an additional layer of mapping to a network slice 882 instance ID. The network slice instance ID further maps to a group 883 of performance requirements and OAM profiles, based on which the 884 network elements within the slice can make local decisions. However, 885 per flow level identification could have adverse impact on the scale 886 of the forwarding entries in the routers. 888 With traditional IP/MPLS VPNs, the set of Route Targets configured 889 for the VPN can be used as some sort of identifier of the VPN in the 890 control plane, and in the data plane, the VPN service labels can be 891 used to identify the data packets belonging to a particular VPN. 892 NVO3 uses the Virtual Network Identifiers (VNIs) in the header of 893 data packets to identify different overlay network tenants. However, 894 It is not clear if the existing identifiers can meet the requirements 895 of network slicing in terms of making local decisions on forwarding 896 policy, QoS and OAM mechanisms, etc. 898 8. Summary 900 The following table is a summary of the identified gaps based on 901 previous analysis in this document. 903 +--------------------------------+---------------------------------------+ 904 | Requirements | Gaps | 905 +--------------------------------+---------------------------------------+ 906 | | | 907 |Req 1. Network Slicing | 1) A detailed specification of NSS | 908 | Specification | | 909 | (NSS) | 2) A companion YANG data model for | 910 | | NSS | 911 +--------------------------------+---------------------------------------+ 912 | | | 913 |Req 2. Network Slicing Cross- | | 914 | Domain Coordination | 3) A companion data model for | 915 | (NS-CDC) | NS-CDC | 916 | | | 917 +--------------------------------+---------------------------------------+ 918 | | | 919 |Req 3. Network Slicing | 4) Slicing specific extension on | 920 | Performance Guarantee and| existing technologies | 921 | Isolation (NS-PGI) | | 922 | | | 923 +--------------------------------+---------------------------------------+ 924 | | | 925 | | 5) Mechanisms for dynamic discovery | 926 | | of service function instances and | 927 | | their capabilities. Mechanisms for| 928 |Req 4. Network Slicing OAM | dynamic discovery of instantiated | 929 | (NS-OAM) | network slices | 930 | | | 931 | | 6) non-overlay OAM solution | 932 | | | 933 | | 7) Mechanisms for customized | 934 | | granularity OAM | 935 | | | 936 +--------------------------------+---------------------------------------+ 938 Table 2: Summary of Gaps 940 9. Security Considerations 942 This document analyzes the standardization work on network slicing in 943 different WGs. As no solution proposed in this document, no security 944 concern raised. 946 10. IANA Considerations 948 There is no IANA action required by this document. 950 11. Acknowledgements 952 The authors wish to thank Hannu Flinck, Akbar Rahman, Ravi Ravindran, 953 Xavier de Foy, Young Lee and Igor Bryskin for their detailed and 954 constructive reviews. Many thanks to Mohamed Boucadair, Christian 955 Jacquenet and Stewart Bryant for their valuable contributions and 956 comments. 958 12. References 960 12.1. Normative References 962 [NS-Framework] 963 "NS Framework", . 966 [NS-UseCase] 967 "NS Use Case", . 970 12.2. Informative References 972 [ANI] "A Reference Model for Autonomic Networking", 973 . 976 [ANIMA-GRASP] 977 "A Generic Autonomic Signaling Protocol (GRASP)", 978 . 981 [DETNET-WG] 982 "Deterministic Networking", 983 . 985 [FGEx] "5G Exchange (5GEx) - Multi-domain Orchestration for 986 Software Defined Infrastructures", 987 . 991 [FLEXE-1.0] 992 "Flexible Ethernet 1.0", . 995 [FlexE-FWK] 996 "FlexE-FWK", . 999 [I-D.boucadair-connectivity-provisioning-protocol] 1000 Boucadair, M., Jacquenet, C., Zhang, D., and P. 1001 Georgatsos, "Connectivity Provisioning Negotiation 1002 Protocol (CPNP)", draft-boucadair-connectivity- 1003 provisioning-protocol-14 (work in progress), May 2017. 1005 [I-D.ietf-spring-segment-routing] 1006 Filsfils, C., Previdi, S., Decraene, B., Litkowski, S., 1007 and R. Shakir, "Segment Routing Architecture", draft-ietf- 1008 spring-segment-routing-12 (work in progress), June 2017. 1010 [I-D.ooamdt-rtgwg-ooam-header] 1011 Mirsky, G., Kumar, N., Kumar, D., Chen, M., Yizhou, L., 1012 and D. Dolson, "OAM Header for use in Overlay Networks", 1013 draft-ooamdt-rtgwg-ooam-header-03 (work in progress), 1014 March 2017. 1016 [I-D.wu-opsawg-service-model-explained] 1017 Wu, Q., LIU, W., and A. Farrel, "Service Models 1018 Explained", draft-wu-opsawg-service-model-explained-06 1019 (work in progress), May 2017. 1021 [I2RS-Yang] 1022 "A Data Model for Network Topologies", 1023 . 1026 [IMT2020-2015] 1027 "Report on Gap Analysis", . 1030 [IMT2020-2016] 1031 "Draft Technical Report Application of network 1032 softwarization to IMT-2020 (O-041)", 1033 . 1036 [IMT2020-2016bis] 1037 "Draft Terms and definitions for IMT-2020 in ITU-T 1038 (O-040)", . 1041 [NGMN-2016] 1042 "Description of Network Slicing Concept", 1043 . 1046 [NMDA] "Network Management Datastore Architecture", 1047 . 1050 [NVO3-WG] "Network Virtualization Overlays". 1052 [ONF-2016] 1053 TS, "Applying SDN Architecture to 5G Slicing", 1054 . 1058 [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1059 Delay Metric for IPPM", RFC 2679, DOI 10.17487/RFC2679, 1060 September 1999, . 1062 [RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1063 Packet Loss Metric for IPPM", RFC 2680, 1064 DOI 10.17487/RFC2680, September 1999, 1065 . 1067 [RFC3209] Awduche, D., Berger, L., Gan, D., Li, T., Srinivasan, V., 1068 and G. Swallow, "RSVP-TE: Extensions to RSVP for LSP 1069 Tunnels", RFC 3209, DOI 10.17487/RFC3209, December 2001, 1070 . 1072 [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation 1073 Metric for IP Performance Metrics (IPPM)", RFC 3393, 1074 DOI 10.17487/RFC3393, November 2002, 1075 . 1077 [RFC4364] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 1078 Networks (VPNs)", RFC 4364, DOI 10.17487/RFC4364, February 1079 2006, . 1081 [RFC4655] Farrel, A., Vasseur, J., and J. Ash, "A Path Computation 1082 Element (PCE)-Based Architecture", RFC 4655, 1083 DOI 10.17487/RFC4655, August 2006, 1084 . 1086 [RFC4664] Andersson, L., Ed. and E. Rosen, Ed., "Framework for Layer 1087 2 Virtual Private Networks (L2VPNs)", RFC 4664, 1088 DOI 10.17487/RFC4664, September 2006, 1089 . 1091 [RFC5440] Vasseur, JP., Ed. and JL. Le Roux, Ed., "Path Computation 1092 Element (PCE) Communication Protocol (PCEP)", RFC 5440, 1093 DOI 10.17487/RFC5440, March 2009, 1094 . 1096 [RFC6291] Andersson, L., van Helvoort, H., Bonica, R., Romascanu, 1097 D., and S. Mansfield, "Guidelines for the Use of the "OAM" 1098 Acronym in the IETF", BCP 161, RFC 6291, 1099 DOI 10.17487/RFC6291, June 2011, 1100 . 1102 [RFC7276] Mizrahi, T., Sprecher, N., Bellagamba, E., and Y. 1103 Weingarten, "An Overview of Operations, Administration, 1104 and Maintenance (OAM) Tools", RFC 7276, 1105 DOI 10.17487/RFC7276, June 2014, 1106 . 1108 [RFC7297] Boucadair, M., Jacquenet, C., and N. Wang, "IP 1109 Connectivity Provisioning Profile (CPP)", RFC 7297, 1110 DOI 10.17487/RFC7297, July 2014, 1111 . 1113 [RFC7432] Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., 1114 Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based 1115 Ethernet VPN", RFC 7432, DOI 10.17487/RFC7432, February 1116 2015, . 1118 [SFCWG] "\Service Function Chaining (sfc)", 1119 . 1121 [SLA-Exchange] 1122 "Inter-domain SLA Exchange Attribute", 1123 . 1126 [TE-Yang] "YANG Data Model for TE Topologies", 1127 . 1130 [TEAS-ACTN] 1131 "Information Model for Abstraction and Control of TE 1132 Networks (ACTN)", . 1135 [TS23-501] 1136 "System Architecture for the 5G System", 1137 . 1140 Authors' Addresses 1142 Li Qiang (editor) 1143 Huawei 1145 Email: qiangli3@huawei.com 1147 Pedro Martinez-Julia 1148 NICT 1150 Email: pedro@nict.go.jp 1152 Liang Geng 1153 China Mobile 1155 Email: gengliang@chinamobile.com 1157 Jie Dong 1158 Huawei 1160 Email: jie.dong@huawei.com 1162 Kiran Makhijani 1163 Huawei 1165 Email: Kiran.Makhijani@huawei.com 1167 Alex Galis 1168 University College London 1170 Email: a.galis@ucl.ac.uk 1172 Susan Hares 1173 Hickory Hill Consulting 1175 Email: shares@ndzh.com 1177 Slawomir 1178 Orange 1180 Email: slawomir.kuklinski@orange.com