| Network Working Group | F. Jounay, Ed. |
| Internet-Draft | Orange CH |
| Intended status: Informational | Y. Kamite, Ed. |
| Expires: April 24, 2014 | NTT Communications |
| G. Heron | |
| Cisco Systems | |
| M. Bocci | |
| Alcatel-Lucent | |
| October 21, 2013 |
Requirements and Framework for Point-to-Multipoint Pseudowires over MPLS PSNs
draft-ietf-pwe3-p2mp-pw-requirements-06.txt
This document presents a set of requirements and a framework for providing a Point-to-Multipoint Pseudowire (PW) over MPLS PSNs. The requirements identified in this document are related to architecture, signaling and maintenance aspects of Point-to-Multipoint PW operation. They are proposed as guidelines for the standardization of such mechanisms. Among other potential applications, Point-to- Multipoint PWs can be used to optimize the support of multicast layer 2 services (Virtual Private LAN Service and Virtual Private Multicast Service) as defined in the Layer 2 Virtual Private Network Working Group.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
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."
This Internet-Draft will expire on April 24, 2014.
Copyright (c) 2013 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.
As defined in the pseudowire architecture [RFC3985], a Pseudowire (PW) is a mechanism that emulates the essential attributes of a telecommunications service (such as a T1 leased line or Frame Relay) over an IP or MPLS PSN. It provides a single service which is perceived by its user as an unshared link or circuit of the chosen service. A Pseudowire is used to transport layer 1 or layer 2 traffic (e.g. Ethernet, TDM, ATM, and FR) over a layer 3 PSN. PWE3 operates "edge to edge" to provide the required connectivity between the two endpoints of the PW.
The Point-to-Multipoint (P2MP) topology described in [I-D.ietf-l2vpn-vpms-frmwk-requirements] and required to provide P2MP L2VPN services can be achieved using one or more P2MP PWs. The use of PW encapsulation enables P2MP services transporting layer 1 or layer 2 data. This could be achieved using a set of point to point PWs, with traffic replication on the PE, but at the cost of bandwidth efficiency, as duplicate traffic would be carried multiple times on shared links.
This document defines the requirements for a Point-to-Multipoint PW (P2MP PW). A P2MP PW is a mechanism that emulates the essential attributes of a P2MP Telecommunications service such as a P2MP ATM VC over a PSN. The required functions of P2MP PWs include encapsulating service-specific PDUs arriving at an ingress Attachment Circuit (AC), and carrying them across a tunnel to one or more egress ACs, managing their timing and order, and any other operations required to emulate the behavior and characteristics of the service as faithfully as possible.
P2MP PWs therefore extend the PWE3 architecture [RFC3985] to offer a P2MP Telecommunications service.
This document also defines the associated requirements related to the P2MP PW operation (e.g. setup and maintenance, protection and scalability).
The document describes the P2MP PW Reference Model architectures and outlines specific signaling requirements for the set up and maintenance of a P2MP PW. In this document, the requirements focus on the Single-Segment PW model. It is for further study how it should be realized in Multi-Segment PW model. For other aspects of P2MP PW implementation, such as packet processing (section 4) and Faithfulness of Emulated Services (section 7), the document refers to [RFC3916].
Some P2MP PW requirements are derived from the signaling requirements for P2MP Traffic-Engineered MPLS Label Switched Paths [RFC4461].
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119] .
This document uses terminology described in [RFC5659]. It also introduces additional terms needed in the context of P2MP PW.
As per the definition of [RFC3985], a pseudowire (PW) both originates and terminates on the edge of the same packet switched network (PSN). The PW label is unchanged between the originating and terminating provider edges (PEs). This is also known as a single-segment pseudowire (SS-PW), as the most fundamental network model of PWE3.
P2MP PW can be defined as Point-to-Multipoint connectivity from a Root PE connected to a traffic source CE to one or more Leaf PEs connected to traffic receiver CEs. It is considered to be an extended architecture of the existing unicast-based SS-PW technology.
Figure 1 describes the P2MP reference model which is derived from [RFC3985] to support P2MP emulated services.
|<-----------P2MP PW -------------->|
Native | | Native
Service | |<----P2MP PSN tunnel --->| | Service
(AC) V V V V (AC)
| +----+ +-----+ +----+ |
| |PE1 | | P |=========|PE2 |AC2 | +----+
| | | | ......PW1.......>|---------->|CE2 |
| | | | . |=========| | | +----+
| | | | . | +----+ |
| | |=========| . | |
| | | | . | +----+ |
+----+ | AC1 | | | . |=========|PE3 |AC3 | +----+
|CE1 |-------->|........PW1.............PW1.......>|---------->|CE3 |
+----+ | | | | . |=========| | | +----+
| | | | . | +----+ |
| | |=========| . | |
| | | | . | +----+ |
| | | | . |=========|PE4 |AC4 | +----+
| | | | ......PW1.......>|---------->|CE4 |
| | | | |=========| | | +----+
| +----+ +-----+ +----+ |
Figure 1: P2MP PW Reference Model
This architecture applies to the case where a P2MP PSN tunnel extends between edge nodes of a single PSN domain to transport a unidirectional P2MP PW with endpoints at these edge nodes. In this model a single copy of each PW packet is sent over the PW on the P2MP PSN tunnel and is received by all Leaf PEs due to the P2MP nature of the PSN tunnel. The P2MP PW MUST be traffic optimized, i.e., only one copy of a P2MP PW packet is sent on any single link. P Routers participate in P2MP PSN tunnel operation but not in the signaling of P2MP PWs.
The Reference Model outlines the basic pieces of a P2MP PW. However, several levels of replication needs to be considered when designing a P2MP PW solution:
Theoretically, it is also possible to consider Ingress PE replication in the core; that is, all traffic is replicated to a set of P2P PSN transport tunnels at ingress, not using P router replication at all. However, this approach may easily lead to more than one-stream bandwidth consumption at a single link, particularly if the PSN tunnels logically go over the same physical link. Hence this approach is not preferred.
Specific operations that must be performed at the PE on the native data units are not described here since the required pre-processing (Forwarder (FWRD) and Native Service Processing (NSP)) defined in section 4.2 of [RFC3985] are also applicable to P2MP PW.
P2MP PWs are generally unidirectional, but a Root PE may need to receive unidirectional P2P traffic from any Leaf PE. For that purpose the P2MP PW solution MAY support optional bidirectional connectivity between the Root PE and each Leaf PE:
Depending on the service using the P2MP PW, the Root PE MAY benefit from information sent by a Leaf PE using P2P connectivity at the expense of the amount of state and configuration overhead for the P2P return path. However, in most situations a Multipoint-to- point (MP2P) connectivity is expected to be sufficient. Hence it MUST be possible for the operator to configure the attributes (P2P or MP2P) of the return path.
The definition of MPLS multicast encapsulation [RFC5332] specifies the procedure to carry MPLS packets that are to be replicated and a copy of the packet sent to each of the specified next hops. This notion is also applicable to P2MP PW (as a MPLS) packet carried by P2MP PSN tunnel.
To be more precise, a P2MP PSN tunnel corresponds to "point-to-multipoint data link or tunnel" described in [RFC5332] Section 3. Similarly, P2MP PW labels correspond to "the top labels (before applying the data link or tunnel encapsulation) of all MPLS packets that are transmitted on a particular point-to-multipoint data link or tunnel."
In P2MP PW architecture, PW label with PW-PDU is replicated by underlying P2MP PSN tunnel layer in SS-PW network model. In other words, it is intended to utilize PSN technology designed for efficient multicast/broadcast trasnport. Note that PW label is unchanged and hidden in transit P routers as long as the model of SS-PW is taken.
In a solution, a P2MP PW MUST be supported over a single P2MP PSN tunnel as underlying layer of traffic distribution. Figure 2 gives an example of P2MP SS-PW topology relying on a single P2MP LSP. The PW tree is composed of one Root PE (i1) and several Leaf PEs (e1, e2, e3, e4).
The mechanisms for establishing the PSN tunnel are outside the scope of this document, as long as they enable the essential attributes of the service to be emulated.
i1
/
/ \
/ \
/ \
/\ \
/ \ \
/ \ \
/ \ / \
e1 e2 e3 e4
Figure 2: Example of P2MP Underlying Layer for P2MP SS-PW
A single P2MP PSN tunnel MUST be able to serve more than one P2MP PW traffic in an aggregated way, i.e., multiplexing.
A P2MP PW solution MAY support different P2MP PSN tunneling technology (e.g., MPLS over GRE [RFC4023], or P-to-MP MPLS LSP) or different setup protocols. (e.g., MLDP [RFC6388], and P2MP RSVP-TE [RFC4875]).
The P2MP LSP associated to the P2MP PW can be selected either by user configuration or by dynamically using a multiplexing/demultiplexing mechanism.
The P2MP PW multiplexing SHOULD be used based on the overlap rate between P2MP LSP and P2MP PW. As an example, an existing P2MP LSP may attach more leaves than the ones defined as Leaf PEs for a given P2MP PW. It may be attractive to reuse it to minimize new configuration, but using this P2MP LSP would imply non-Leaf PEs receive unwanted traffic, not destined to Leaf PE at the service layer. The operator should determine whether the P2MP PW can accept partially multiplexing with P2MP LSP, and a minimum congruency rate may be defined. The Root PE can determine whether P2MP PW can multiplex to a P2MP LSP according to the congruency rate. The congruency rate should take into account several items, such as:
With this procedure a P2MP PW is nested within a P2MP LSP. This allows multiplexing several PWs over a common P2MP LSP. Prior to the P2MP PW signaling phase, the Root PE determines which P2MP LSP will be used for this P2MP PW. The PSN Tunnel can be an existing PSN tunnel or the Root PE can create a new P2MP PSN tunnel.
[RFC5332] introduces two approaches to assign MPLS label (meaning PW label in P2MP PW's context): Upstream-Assigned[RFC5331] and Downstream-Assigned. However, it is out of scope of this document which one should be used in PW construction. It is left to the specification of the solution work.
The following requirements apply to the establishment of P2MP PWs (P2MP SS-PWs):
The Root PE SHOULD support a method to be informed about whether a Leaf PE has successfully attached to the PW tree.
The P2MP PW MUST be uniquely identified. This unique P2MP PW identifier MUST be used for all signaling procedures related to this PW (PW setup, monitoring, etc).
The Root PE and Leaf PEs of a P2MP PW MUST be configured with the same PW type as defined in [RFC4446] for P2P PW. In case of a different type, a PE MUST abort attempts to establish the P2MP PW.
Some interface parameters [RFC4446] related to the AC capability have been defined according to the PW type and are signaled during the PW setup.
Where applicable, a solution is REQUIRED to ascertain whether the AC at the Leaf PE is capable of supporting traffic coming from the AC at the Root PE.
In case of a mismatch, the passive PE (Root or Leaf PE, depending on the signaling process) MUST support mechanisms to reject attempts to establish the P2MP SS-PW.
Once the PW tree is established, the solution MUST allow the addition or removal of a Leaf PE, or a subset of leaves to/from the existing tree, without any impact on the PW tree (data and control planes) for the remaining Leaf PEs.
The addition or removal of a Leaf PE MUST also allow the P2MP PSN tunnel to be updated accordingly. This may cause the P2MP PSN tunnel to add or remove the corresponding Leaf PE.
Since the underlying layer has an End-to-End P2MP topology between the Root PE and the Leaf PEs, the failure reporting and processing procedures are implemented only on the edge nodes.
Failure events may cause one or more Leaf PEs to become detached from the PW tree. These events MUST be reported to the Root PE, using appropriate out-of-band or inband OAM messages.
It MUST be possible for the operator to choose the out-of-band or inband OAM tools or both to monitor the Leaf PE status. The solution SHOULD allow the Root PE to be informed of Leaf PEs failure for management purposes.
Based on these failure notifications, solutions MUST allow the Root PE to update the remaining leaves of the PW tree.
It is assumed that if recovery procedures are required, the P2MP PSN tunnel will support standard MPLS-based recovery techniques (typically based on RSVP-TE). In that case a mechanism SHOULD be implemented to avoid race conditions between recovery at the PSN level and recovery at the PW level.
An alternative protection scheme MAY rely on the PW layer.
Leaf PEs MAY be protected via a P2MP PW redundancy mechanism. In the example depicted below, a standby P2MP PW is used to protect the active P2MP. In that protection scheme the AC at the Root PE MUST serve both P2MP PWs. In this scenario, the condition when to do the switchover SHOULD be implemented, e.g. one or all Leaf failure of active P2MP PW will course P2MP PW switchover.
CE1
|
active PE1 standby
P2MP PW .../ \....P2MP PW
/ \
P2 P3
/ \ / \
/ \ / \
/ \ / \
PE4 PE5 PE6 PE7
| | | |
| \ / |
\ CE2 /
\ /
-------CE3------
Figure 3: Example of P2MP PW redundancy for protecting Leaf PEs
The Root PE MAY be protected via a P2MP PW redundancy mechanism. In the example depicted below, a standby P2MP PW is used to protect the active P2MP. A single AC at the Leaf PE MUST be used to attach the CE to the primary and the standby P2MP PW. The Leaf PE MUST support protection mechanisms in order to select the active P2MP PW.
CE1
/ \
| |
active PE1 PE2 standby
P2MP PW1 | | P2MP PW2
| |
P2 P3
/ \/ \
/ /\ \
/ / \ \
/ / \ \
PE4 PE5
| |
CE2 CE3
Figure 4: Example of P2MP PW redundancy for protecting Root PEs
The solution SHOULD scale at least linearly with the number of Leaf PEs.
Increasing the number of P2MP PWs between a Root PE and a given set of Leaf PEs SHOULD NOT cause the P router to increase the number of entries in its forwarding table by the same or greater proportion. Multiplexing P2MP PWs to P2MP PSN Tunnels achieves this.
The solution SHOULD provide a simple provisioning procedure to build a P2MP PW.
The solution MUST take into consideration the situation where the Root PE and Leaf PEs are not managed by a single NMS.
In that case it MUST be possible to manage the whole P2MP PW using a single NMS. Typically the P2MP PW could be managed from the Root PE.
Solutions MUST be backward compatible with current PW standards. Solutions SHOULD utilize existing capability advertisement and negotiation procedures for the PEs implementing P2MP PW endpoints.
The implementation of OAM mechanisms also implies the advertisement of PE capabilities to support specific OAM features. The solution MAY allow advertising P2MP PW OAM capabilities.
A solution MUST NOT allow a P2PW to be established to PEs that do not support P2MP PW functionality. It MUST have a mechanism to report an error for incompatible PEs.
In some cases, upstream traffic is needed from downstream CEs to upstream CEs. The P2MP PW solution SHOULD allow a return path (i.e. from the Leaf to the Root) that provides upstream connectivity.
In particular, the same ACs MAY be shared between downstream and upstream directions. For downstream, a CE receives traffic originated by the Root PE over its AC. For upstream, the CE MAY also send traffic destined to the same Root PE over the same AC.
The security requirements common to PW are raised in Section 10 of [RFC3916]. P2MP PW is a variant of the initial P2P PW definition, and those requirements also apply to P2MP PW.
This draft does not require any IANA action.
Philippe Niger
France Telecom
2, avenue Pierre-Marzin
22307 Lannion Cedex
France
Email: philippe.niger@orange-ftgroup.com
Luca Martini
Cisco Systems, Inc.
9155 East Nichols Avenue, Suite 400
Englewood, CO, 80112
EMail: lmartini@cisco.com
Lei Wang
Telenor
Snaroyveien 30
Fornebu 1331
Norway
Email: lei.wang@telenor.com
Rahul Aggarwal
Juniper Networks
1194 North Mathilda Ave.
Sunnyvale, CA 94089
Email: rahul@juniper.net
Simon Delord
Alcatel-Lucent
Building 3, 388 Ningqiao Road, Jinqiao, Pudong
Shanghai, 201206, P.R. China
Email: simon.delord@alcatel-lucent.com
Martin Vigoureux
Alcatel-Lucent France
Route de Villejust
91620 Nozay
France
Email: martin.vigoureux@alcatel-lucent.fr
Lizhong Jin
ZTE Corporation
889, Bibo Road,
Shanghai, 201203, China
Email: lizhong.jin@zte.com.cn
The authors thank the authors of [RFC4461] since the structure and content of this document were, for some sections, largely inspired by [RFC4461]. Many thanks to JL Le Roux and A. Cauvin for the discussions, comments and support.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. |