TEAS WG Young Lee Internet Draft Dhruv Dhody Intended status: Informational Huawei Daniele Ceccarelli Ericsson Oscar Gonzalez de Dios Telefonica Expires: September 2017 March 13, 2017 Abstraction and Control of TE Networks (ACTN) Abstraction Methods draft-lee-teas-actn-abstraction-01 Status of this Memo This Internet-Draft is submitted to IETF in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on September 13, 2017. Copyright Notice Lee, et al. Expires September 13, 2017 [Page 1] Internet-Draft ACTN Abstraction Methods March 2017 Copyright (c) 2016 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Abstract Abstraction and Control of Traffic Engineering (TE) Networks(ACTN) refers to the set of virtual network operations needed to orchestrate, control and manage large-scale multi-domain TE networks, so as to facilitate network programmability, automation, and efficient resource sharing. As the ACTN architecture considers abstraction as one of the important building blocks, this document describes a few alternatives methods of abstraction for both packet and optical networks. This is an important consideration since the choice of the abstraction method impacts protocol design and the information it carries. Table of Contents 1. Introduction...................................................3 2. ACTN Architecture..............................................4 3. Abstraction Factors and Methods................................5 3.1. No abstraction (native/white topology)....................6 3.2. One Abstract Node (black topology)........................7 3.3. Abstraction of TE tunnels for all pairs of border nodes (grey topology)................................................9 3.3.1. Grey topology type A: border nodes with a TE links between them in a full mesh fashion........................10 3.3.2. Grey topology Type B................................11 3.4. How to build grey topology...............................11 Lee, et. al. Expires September 13, 2017 [Page 2] Internet-Draft ACTN Abstraction Methods March 2017 3.4.1. Automatic generation of abstract topology by configuration..............................................11 3.4.2. On-demand generation of supplementary topology via path compute request/reply......................................12 4. Protocol/Data Model Requirements..............................13 4.1. Packet Networks..........................................13 4.2. OTN Networks.............................................14 4.3. WSON Networks............................................14 5. Security......................................................14 6. Acknowledgements..............................................15 7. References....................................................15 7.1. Informative References...................................15 8. Contributors..................................................15 Authors' Addresses...............................................16 Appendix A:......................................................16 1. Introduction Abstraction and Control of TE Networks (ACTN) describes a method for operating a Traffic Engineered (TE) network (such as an MPLS-TE network or a layer 1 transport network) to provide connectivity and virtual network services for customers of the TE network. The services provided can be tuned to meet the requirements (such as traffic patterns, quality, and reliability) of the applications hosted by the customers. More details about ACTN can be found in Section 2. Abstraction is defined in [RFC7926] as: Abstraction is the process of applying policy to the available TE information within a domain, to produce selective information that represents the potential ability to connect across the domain. Thus, abstraction does not necessarily offer all possible connectivity options, but presents a general view of potential connectivity according to the policies that determine how the domain's administrator wants to allow the domain resources to be used. Connectivity referred to this document is TE path through a series of connected domains as used in [RFC7926]. As the ACTN architecture considers abstraction as one of the important building blocks, this document discusses a few alternatives for the methods of abstraction for both packet and optical networks. This is an important consideration since the choice of the abstraction method impacts protocol design and the information it carries. Lee, et. al. Expires September 13, 2017 [Page 3] Internet-Draft ACTN Abstraction Methods March 2017 The purpose of this document is to find a common agreement on the factors and methods of abstraction. These abstraction factors and methods may in turn impact implementations and protocol design. 2. ACTN Architecture This section provides a brief description of ACTN architecture. [ACTN-Frame] describes the architecture model for ACTN including the entities (Customer Network Controller (CNC), Multi-domain Service Coordinator (MDSC), and Physical Network Controller (PNC) and their interfaces. Figure 1 depicts a high-level control and interface architecture for ACTN and is a reproduction of Figure 5 from [ACTN-Frame]. VPN customer NW Mobile Customer ISP NW service Customer | | | +-------+ +-------+ +-------+ | CNC-A | | CNC-B | | CNC-C | +-------+ +-------+ +-------+ \ | / ----------- |CMI I/F -------------- \ | / +-----------------------+ | MDSC | +-----------------------+ / | \ ------------- |MPI I/F ------------- / | \ +-------+ +-------+ +-------+ | PNC | | PNC | | PNC | +-------+ +-------+ +-------+ | GMPLS / | / \ | trigger / | / \ -------- ---- +-----+ +-----+ \ ( ) ( ) | PNC | | PCE | \ - - ( Phys ) +-----+ +-----+ ----- ( GMPLS ) (Netw) | / ( ) ( Physical ) ---- | / ( Phys. ) ( Network ) ----- ----- ( Net ) - - ( ) ( ) ----- ( ) ( Phys. ) ( Phys ) -------- ( Net ) ( Net ) ----- ----- Figure 1 : ACTN Control Hierarchy Lee, et. al. Expires September 13, 2017 [Page 4] Internet-Draft ACTN Abstraction Methods March 2017 The MDSC oversees the specific aspects of the different domains and builds a single abstracted end-to-end network topology in order to coordinate end-to-end path computation and path/service provisioning. In order for the MDSC to perform its coordination function, it depends on the coordination with the PNCs which are the domain-level controllers especially as to what level of domain network resource abstraction is agreed upon between the MDSC and the PNCs. As discussed in [RFC7926], abstraction is tied with policy of the networks. For instance, per an operational policy, the PNC would not be allowed to provide any technology specific details (e.g., optical parameters for WSON) in its update. In such case, the abstraction level of the update will be in a generic nature. In order for the MDSC to get technology specific topology information from the PNC, a request/reply mechanism may be employed. In some cases, abstraction is also tied with the controller's capability of abstraction as abstraction involves some rules and algorithms to be applied to the actual network resource information (which is also known as network topology). [TE-Topology] describes YANG models for TE-network abstraction. [PCEP-LS] describes PCEP Link-state mechanism that also allows for transport of abstract topology in the context of Hierarchical PCE. 3. Abstraction Factors and Methods This section discusses factors that may impact the choice of abstraction and presents a number of abstraction methods. It is important to understand that abstraction depends on several factors: - The nature of underlying domain networks: Abstraction depends on the nature of the underlying domain networks. For instance, packet networks may have different level of abstraction requirements from that of optical networks. Within optical networks, WSON may have different level of abstraction requirements than the OTN networks. - The capability of the PNC: Abstraction depends on the capability of the PNCs. As abstraction requires hiding details of the underlying resource network resource information, the PNC capability to run some internal optimization algorithm impacts the feasibility of abstraction. Some PNC may not have the ability to Lee, et. al. Expires September 13, 2017 [Page 5] Internet-Draft ACTN Abstraction Methods March 2017 abstract native topology while other PNCs may have such an ability to abstract actual topology by using sophisticated algorithms. - Scalability factor: Abstraction is a function of scalability. If the actual network resource information is of small size, then the need for abstraction would be less than the case where the native network resource information is of large size. In some cases, abstraction may not be needed at all. - The frequency of topology updates: The proper abstraction level may depend on the frequency of topology updates and vice versa. - The capability/nature of the MDSC: The nature of the MDSC impacts the degree/level of abstraction. If the MDSC is not capable of handling optical parameters such as those specific to OTN/WSON, then white topology abstraction may not work well. - The confidentiality: In some cases where the PNC would like to hide key internal topological data from the MDSC, the abstraction method should consider this aspect. - The scope of abstraction: All of the aforementioned factors are equally applicable to both the MPI (MDSC-PNC Interface) and the CMI (CNC-MDSC Interface). With having the aforementioned factors in mind, the following abstraction methods can be considered for implementations. 3.1. No abstraction (native/white topology) This is a case where the PNC provides the actual network topology to the MDSC without any hiding or filtering. In this case, the MDSC has the full knowledge of the underlying network topology and as such there is no need for the MDSC to send a path computation request to the PNC. The computation burden will fall on the MDSC to find an optimal end-to-end path and optimal per domain paths. Lee, et. al. Expires September 13, 2017 [Page 6] Internet-Draft ACTN Abstraction Methods March 2017 +--+ +--+ +--+ +--+ +-+ +-----+ +-----+ +-----+ +-+ ++-+ ++-+ +-++ +-++ | | | | | | | | | | | | | | | | ++-+ ++-+ +-++ +-++ +-+ +-----+ +-----+ +-----+ +-+ +--+ +--+ +--+ +--+ Figure 1: The native/white topology 3.2. One Virtual Node (black topology) The entire domain network is abstracted as a single virtual node (see the definition of virtual node in [RFC7926]) with the access/egress links without disclosing any node internal connectivity information. Figure 2a depicts a native topology with the corresponding black topology with one virtual node and inter-domain links. In this case, the MDSC has to make path computation requests to the PNCs before it can determine an end-to-end path. If there are a large number of inter-connected domains, this abstraction method may impose a heavy coordination load at the MDSC level in order to find an optimal end- to-end path. Figure 2b depicts another type of a black topology with border nodes and inter-domain links. The black topology would not give the MDSC any critical network resource information other than the border nodes/links information and as such it is likely to have a need for complementary communications between the MDSC and the PNCs (e.g., Path Computation Request/Reply). Lee, et. al. Expires September 13, 2017 [Page 7] Internet-Draft ACTN Abstraction Methods March 2017 +--+ +--+ +--+ +--+ +-+ +-----+ +-----+ +-----+ +-+ ++-+ ++-+ +-++ +-++ | | | | | | | | | | | | | | | | ++-+ ++-+ +-++ +-++ +-+ +-----+ +-----+ +-----+ +-+ +--+ +--+ +--+ +--+ +--------+ +--+ +--+ | | | | | | | | | | | | +--+ +--+ +--------+ Figure 2a: The native topology and the corresponding black topology with one virtual node and inter-domain links ----- ( ) ( ) +--+ +--+ +-+ | | +-+ +--+ +--+ ( ) ( ) ( ) +--+ +--+ +-+ | | +-+ +--+ +--+ ( ) ( ) ----- Lee, et. al. Expires September 13, 2017 [Page 8] Internet-Draft ACTN Abstraction Methods March 2017 Figure 2b: A black topology with border nodes and inter-domain links 3.3. Abstraction of TE tunnels for all pairs of border nodes (grey topology) This abstraction level, referred to a grey topology in [ACTN-frame] is between black topology and white topology from a granularity point of view. As shown in Figures 3a and 3b, we may further differentiate from a perspective of how to abstract internal TE resources between the pairs of border nodes: . Grey topology type A: border nodes with a TE links between them in a full mesh fashion (See Figure 3a) . Grey topology type B: border nodes with some internal abstracted nodes and abstracted links (See Figure 3b) +--+ +--+ +--+ +--+ +-+ +-----+ +-----+ +-----+ +-+ ++-+ ++-+ +-++ +-++ | | | | | | | | | | | | | | | | ++-+ ++-+ +-++ +-++ +-+ +-----+ +-----+ +-----+ +-+ +--+ +--+ +--+ +--+ +--+ +--+ +-+ +----+ +-+ ++-+ +-++ | \ / | | \/ | | /\ | | / \ | ++-+ +-++ +-+ +----+ +-+ +--+ +--+ Lee, et. al. Expires September 13, 2017 [Page 9] Internet-Draft ACTN Abstraction Methods March 2017 Figure 3a: The native topology and the corresponding grey topology type A with TE links between border nodes +--+ +--+ +--+ +-+ +-----+ +-----+ +-+ ++-+ ++-+ +-++ | | | | | | | | ++-+ ++-+ +-++ +-+ +-----+ +-----+ +-+ +--+ +--+ +--+ Figure 3b: The grey topology type B with abstract nodes/links between border nodes 3.3.1. Grey topology type A: border nodes with a TE links between them in a full mesh fashion For each pair of ingress and egress nodes (i.e., border nodes to/from the domain), TE link metric is provided with TE attributes such as max bandwidth available, link delay, etc. This abstraction depends on the underlying TE networks. Note that this topology is similar to the connectivity matrix defined in [TE-Topology]. The only thing might be different is some additional information about the end points of the links of the border nodes if they cannot be included in the connectivity matrix's termination points. - For packet networks, abstraction may include max bandwidth available, delay, etc. - For OTN networks, max bandwidth available may be per ODU 0/1/2/3 switching level or aggregated across all ODU switching levels (i.e., ODUj/k).Clearly, there is a trade-off between these two Lee, et. al. Expires September 13, 2017 [Page 10] Internet-Draft ACTN Abstraction Methods March 2017 abstraction methods. Some OTN switches can switch any level of ODUs and in such case there is no need for ODU level abstraction. - For WSON networks, max bandwidth available may be per lambda/frequency level (OCh) or aggregated across all lambda/frequency level. Per OCh level abstraction gives more detailed data to the MDSC at the expense of more information processing. Either OCh-level or aggregated level abstraction should factor in the RWA constraint (i.e., wavelength continuity) at the PNC level. This means the PNC should have this capability and advertise it as such. See the Appendix for this abstraction method. 3.3.2. Grey topology Type B The grey abstraction type B would allow the MDSC to have more information about the internals of the domain networks by the PNCs so that the MDSC can flexibly determine optimal paths. The MDSC may configure some of the internal virtual nodes (e.g., cross-connect) to redirect its traffic as it sees changes from the domain networks. 3.4. How to build grey topology This section discusses two different methods of building a grey topology: . Automatic generation of abstract topology by configuration (Section 3.4.1) . On-demand generation of supplementary topology via path computation request/reply (Section 3.4.2) 3.4.1. Automatic generation of abstract topology by configuration The "Automatic generation" method is based on the abstraction/summarization of the whole domain by the PNC and its advertisement on MPI interface once the abstraction level is configured. The level of abstraction advertisement can be decided based on some PNC configuration parameters (e.g. provide the potential connectivity between any PE and any ASBR in an MPLS-TE network as described in section 3.3.1) Note that the configuration parameters for this potential topology can include available B/W, latency, or any combination of defined parameters. How to generate such tunnel information is beyond the Lee, et. al. Expires September 13, 2017 [Page 11] Internet-Draft ACTN Abstraction Methods March 2017 scope of this document. Appendix A provides one example of this method for the WSON case. Such potential topology needs to be periodically or incrementally/asynchronously updated every time that a failure, a recovery or the setup of new VNs causes a change in the characteristics of the advertised grey topology (e.g. in our previous case if due to changes in the network is it now possible to provide connectivity between a given PE and a given ASBR with a higher delay in the update). 3.4.2. On-demand generation of supplementary topology via path compute request/reply The "on-demand generation" of supplementary topology is to be distinguished from automatic generation of abstract topology. While abstract topology is generated and updated automatically by configuration as explained in Section 3.4.1., additional supplementary topology may be obtained by the MDSC via path compute request/reply mechanism. Starting with a black topology advertisement from the PNCs, the MDSC may need additional information beyond the level of black topology from the PNCs. It is assumed that the black topology advertisement from PNCs would give the MDSC each domain's the border node/link information as described in Figure 2. Under this scenario, when the MDSC needs to allocate a new VN, the MDSC can issue a number of Path Computation requests as described in [ACTN-YANG] to different PNCs with constraints matching the VN request. An example is provided in Figure 4, where the MDSC is requesting to setup a P2P VN between AP1 and AP2. The MDSC can use two different inter-domain links to get from Domain X to Domain Y, namely the one between ASBRX.1 and ASBRY.1 and the one between ASBRX.2 and ASBRY.2, but in order to choose the best end to end path it needs to know what domain X and Y can offer in term of connectivity and constraints between the PE nodes and the ASBR nodes. ------- ------- ( ) ( ) - ASBRX.1------- ASBRY.1 - (+---+ ) ( +---+) -+---( |PE1| Dom.X ) ( Dom.Y |PE2| )---+- | (+---+ ) ( +---+) | AP1 - ASBRX.2------- ASBRY.2 - AP2 Lee, et. al. Expires September 13, 2017 [Page 12] Internet-Draft ACTN Abstraction Methods March 2017 ( ) ( ) ------- ------- Figure 4: A multi-domain networks example A path computation request will be issued to PNC.X asking for potential connectivity between PE1 and ASBRX.1 and between PE1 and ASBRX.2 with related objective functions and TE metric constraints. A similar request will be issued to PNC.Y and the results merged together at the MDSC to be able to compute the optimal end-to-end path including the inter domain links. The info related to the potential connectivity may be cached by the MDSC for subsequent path computation processes or discarded, but in this case the PNCs are not requested to keep the grey topology updated. 4. Protocol/Data Model Requirements This section provides a set of requirements that may impact the way protocol/data model is designed and the information elements thereof which are carried in the protocol/data model. It is expected that the abstraction level be negotiated between the CNC and the MDSC (i.e., the CMI) depending on the capability of the CNC. This negotiated level of abstraction on the CMI may also impact the way the MDSC and the PNCs configure and encode the abstracted topology. For example, if the CNC is capable of sophisticated technology specific operation, then this would impact the level of abstraction at the MDSC with the PNCs. On the other hand, if the CNC asks for a generic topology abstraction, then the level of abstraction at the MDSC with the PNCs can be less technology specific than the former case. The subsequent sections provide a list of possible abstraction levels for various technology domain networks. 4.1. Packet Networks - For grey abstraction, the type of abstraction and its parameters MUST be defined and configured/negotiated. o Abstraction Level 1: TE-tunnel abstraction for all (S-D) border pairs with: . Maximum B/W available per Priority Level . Minimum Latency Lee, et. al. Expires September 13, 2017 [Page 13] Internet-Draft ACTN Abstraction Methods March 2017 o Other Level (TBD) 4.2. OTN Networks - For grey abstraction, the type of abstraction and its parameters MUST be defined and configured/negotiated. o Abstraction Level 1: Per ODU Switching level (i.e., ODU type and number) TE-tunnel abstraction for all (S-D) border pairs with: . Maximum B/W available per Priority Level . Minimum Latency o Abstraction Level 2: Aggregated TE-tunnel abstraction for all (S-D) border pairs with: . Maximum B/W available per Priority Level . Minimum Latency o Other Level (TBD) 4.3. WSON Networks - For grey abstraction, the type of abstraction MUST and its parameters be defined and configured/negotiated. o Abstraction Level 1: Per Lambda/Frequency level TE-tunnel abstraction for all (S-D) border pairs with: . Maximum B/W available per Priority Level . Minimum Latency o Abstraction Level 2: Aggregated TE-tunnel abstraction for all (S-D) border pairs with: . Maximum B/W available per Priority Level . Minimum Latency o Other Level (TBD) Note that Appendix A provides how to compute WSON grey topology Abstraction Level 1 and Level 2. These examples illustrate that the encoding of an abstraction topology can be impacted by the configured/negotiated abstraction level in the ACTN interfaces. Lee, et. al. Expires September 13, 2017 [Page 14] Internet-Draft ACTN Abstraction Methods March 2017 5. Acknowledgements We thank Name for providing useful comments and suggestions for this draft. 6. References 6.1. Informative References [RFC7926] A. Farrel, Ed., "Problem Statement and Architecture for Information Exchange between Interconnected Traffic- Engineered Networks", RFC 7926, July 2016. [ACTN-Frame] D. Cecarelli and Y. Lee, "Framework for Abstraction and Control of Traffic Engineered Networks", draft-ietf-teas- actn-framework, work in progress. [TE-Topology] X. Liu, et. al., "YANG Data Model for TE Topologies", draft-ietf-teas-yang-te-topo, work in progress. [PCEP-LS] D. Dhody, Y. Lee and D. Ceccarelli, "PCEP Extension for Distribution of Link-State and TE Information," draft- dhodylee-pce-pcep-ls, work in progress. [RFC7926] A. Farrel, et. al., "Problem Statement and Architecture for Information Exchange Between Interconnected Traffic Engineered Networks", RFC 7926, July 2016. [ACTN-YANG] X. Zhang, et. Al., "Applicability of YANG models for Abstraction and Control of Traffic Engineered Networks", draft-zhang-teas-actn-yang, work in progress 7. Contributors Contributor's Addresses Sergio Belotti Nokia Email: sergio.belotti@nokia.com Xian Zhang Huawei Lee, et. al. Expires September 13, 2017 [Page 15] Internet-Draft ACTN Abstraction Methods March 2017 Email: zhang.xian@huawei.com Authors' Addresses Young Lee Huawei Technologies 5340 Legacy Drive Plano, TX 75023, USA Phone: (469)277-5838 Email: leeyoung@huawei.com Dhruv Dhody Huawei Technologies Email: dhruv.ietf@gmail.com Daniele Ceccarelli Ericsson Torshamnsgatan,48 Stockholm, Sweden Email: daniele.ceccarelli@ericsson.com Oscar Gonzalez de Dios Telefonica Email: oscar.gonzalezdedios@telefonica.com Appendix A: This section provides how WSON grey topology abstraction levels 1 and 2 can be computed at a PNC. These examples illustrate that the encoding of an abstraction topology can be impacted by the configured/negotiated abstraction level at the MPI. Abstraction Level 1: Per Lambda/Frequency level TE-tunnel abstraction for all (S-D) border pairs: For each (S-D) border node pair, Lee, et. al. Expires September 13, 2017 [Page 16] Internet-Draft ACTN Abstraction Methods March 2017 1) The concept of a lambda plane: A lambda plane is a confined optical topology with respect to a given lambda value. If an OMS link has the wavelength of the given lambda available, it is included, otherwise excluded. 2) Calculate the maximal flow between S and D in every lambda plane. Max flow computation is restricted to each lambda plane is for OCh wavelength continuity. 3) Convert each feasible lambda plane with OCh wavelength continuity to B/W equivalent encoding; Send this per lambda level encoding for (S-D) to the MDSC; Abstraction Level 2: Aggregated TE-tunnel abstraction for WSON for all (S-D) border pairs For each (S-D) border node pair, 1) The concept of a lambda plane: A lambda plane is a confined optical topology with respect to a given lambda value. If an OMS link has the wavelength of the given lambda available, it is included, otherwise excluded. 2) Calculate the maximal flow between S and D in every lambda plane. Max flow computation is restricted to each lambda plane is for OCh wavelength continuity. 3) Add up the max flow values across all lambda planes. This is the maximal number of OCh paths that can be setup between S and D at the same time. 4) Convert the max number of OCh paths to B/W equivalent encoding; Send this encoding as max B/W for (S-D) to the MDSC; Lee, et. al. Expires September 13, 2017 [Page 17]