Internet-Draft | PCEP p2mp sr policy | July 2022 |
Bidgoli, et al. | Expires 7 January 2023 | [Page] |
SR P2MP policies are set of policies that enable architecture for P2MP service delivery. This document specifies extensions to the Path Computation Element Communication Protocol (PCEP) that allow a stateful PCE to compute and initiate P2MP paths from a Root to a set of Leaves.¶
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 https://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 7 January 2023.¶
Copyright (c) 2022 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 (https://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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
The draft [draft-ietf-pim-sr-p2mp-policy] defines a variant of the SR Policy that uses [draft-ietf-spring-segment-routing-policy] for constructing a P2MP segment to support multicast service delivery.¶
A Point-to-Multipoint (P2MP) Policy connects a Root node to a set of Leaf nodes, optionally through a set of intermediate replication nodes. A Replication segment [draft-ietf-spring-sr-replication-segment], corresponds to the state of a P2MP segment on a particular node and provide forwarding instructions for the segment.¶
A P2MP Policy is relevant on the root of the P2MP Tree and it contains candidate paths. The candidate paths are made of path-instances and each path-instance is constructed via replication segments. These replication segments are programmed on the root, leaves and optionally intermediate replication nodes.¶
Replication segments MAY be connected to each other directly, or they MAY be connected or steered via unicast SR segments or a segment list.¶
For a P2MP Tree, a controller may be used to compute paths from a Root node to a set of Leaf nodes, optionally via a set of replication nodes. A packet is replicated at the root node and optionally on Replication nodes towards each Leaf node.¶
There are two types of a P2MP Tree: Spray and Replication.¶
A Point-to-Multipoint service delivery could be via Ingress Replication, known as Spray. The root unicasts individual copies of traffic to each leaf. The corresponding P2MP Policy consists of replication segments only for the root and the leaves and they are connected via a unicast SR Segment.¶
A Point-to-Multipoint service delivery could also be via Downstream Replication, known as Replication Tree. The root and some downstream replication nodes replicate the traffic along the tree as it traverses closer to the leaves.¶
The PCE discovers the root and the leaves via different methods. As an example, the leaves and the root can be explicitly configured on the PCE or PCC can update the PCE with the identity of the root and the leaves when it discovers them via multicast protocols like MP-BGP and MVPN procedures [RFC6513] or PIM. The controller can calculate the P2MP Policy and any of its associated replication segments from the root to the leaves with these info and any additional Service Leave Agreements (SLAs) that is used to construct the tree.¶
This document defines PCEP objects, TLVs and the procedures to instantiate a P2MP Policy and Replication Segments.¶
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 RFC 2119 [RFC2119].¶
After discovering the root and the leaves the PCE programs the PCCs with relevant information needed to create a P2MP Tree.¶
As per draft [draft-ietf-pim-sr-p2mp-policy] a P2MP Policy is defined by Root-ID, Tree-ID and a set of leaves. A P2MP policy is a variant of SR policy as such it uses the same concept as draft [draft-ietf-pce-segment-routing-policy-cp]. A P2MP policy is composed of a collection of SR P2mp Candidate Paths. Candidate paths are computed by the PCE and can be used for P2MP Tree redundancy. Only a single candidate path may be active at each time. Each candidate paths can be globally optimized, therefore it is consists of multiple path-instances. A path-instance can be considered to a P2MP LSP. If a candidate path needs to be globally optimized two path-instances can be programmed from the root the leaves and via make before break procedures the candidate path can be switched from path-instance 1 to the 2nd path-instance. The forwarding states of these path-instances are build via replication segments, in short each path-instance initiated on the root has its own set of replication segments on the Root, Transit and Leaf nodes.¶
A replication segment is set of forwarding instructions on a specific node. Each instruction may be a PUSH or SWAP operation before forwarding out of an interface, or a POP action on bud and leaf nodes.¶
PCE could also calculate and download additional information for the replication segments, such as protections next-hops for link protection (FRR).¶
SR P2MP Policy¶
Is only relevant on the Root of the P2MP tree and is a policy on PCE. It is downloaded only on the root node and is identified via <Root-ID, Tree-ID> It contains the following information:¶
Candidate Path:¶
Replication Segment:¶
Explained further in upcoming sections, there are 2 ways to identify the replication segment, depending on the type of replication segment (shared replication segment or non-shared replication segment)¶
A non-shared Replication Segment is used when the label field of the PMSI Tunnel Attribute (PTA) is set to zero as per [draft-parekh-bess-mvpn-sr-p2mp]. This is used when there is no upstream assigned label in the PTA (provider tunnel attribute) and aggregate of MVPNs into a single P-Tunnel is not desired.¶
An alternative a Replication Segment MAY be shared by P2MP trees, e.g. for protection. A shared Replication segment MAY be identified with zero Root-ID address (0.0.0.0 for IPv4 and :: for IPv6) and a Replication- ID that is unique in context of Node address where the Replication segment is instantiated and MUST NOT be associated a particular tree or P2MP Policy.¶
This document attempts to leverage existing IETF draft and RFC documents which define PCEP objects, to update the PCE with Root and Leaves information when PCC Initiated method is used. Similarly, existing documents are utilized where feasible to update the PCC with relevant information to build the P2MP Policy and its Replication Segments. This document introduces new TLVs and Objects specific to a programing P2MP policy and its replication segment.¶
[RFC8231] The bases for a stateful PCE, and reuses the following objects or a variant of them¶
It should be noted that the [draft-dhs-spring-sr-p2mp-policy-yang] can provide further details of the high level P2MP Policy Model.¶
A P2MP Policy and its candidate path can be identified on the root via the P2MP LSP Object. This Object is a variation of the LSP ID Object defined in [RFC8231] and is as follow:¶
PLSP-ID: [RFC8231], is assigned by PCC and is unique per candidate path. It is constant for the lifetime of a PCEP session. Stand-by candidate paths will be assigned a new PLSP-ID by PCC. Stand-by candidate paths can co-exist with the active candidate path.¶
Instance-ID: LSP ID Identifier as defined in RFC 3209, is the path-instance identifier and is assigned by the PCC. As it was mentioned the candidate path can have up to two path-instance for global optimization. Note that the Root-ID, Tree-ID and Instance-ID are part of a new SR- P2MP-LSP-IDENTIFIER TLV which will be identified in this draft.¶
The key to identify a replication segment is also a P2MP LSP Object. With varying encoding rules for the SR-P2MP-LSP- IDENTIFIER TLV which will be explained in later sections.¶
PCECC and a variant of CCI object is used in Replication Segment to identify a cross connect. A cross connect is a incoming SID and set of outgoing interfaces and their corresponding SID or SID List. The CCI objects contains the incoming SID and the outgoing interfaces which are presented via the ERO objects, which each may contain a list of segments.¶
A P2MP policy can be instantiated via the PCC or the PCE depending on how the root and the leaves are discovered. This document describes two way to discover the root and the leaves:¶
Root in response to the PCInitiate message, will generate PLSP-ID for the candidate paths and an Instance-ID for the Path-Instance (LSP-ID) contained with in the candidate path. The tree-id for the P2MP Policy is also filled. PCC will reports back the PLSP-ID, Instance-ID and tree-id via PCRpt message¶
After Root (PCC) discovers the leaves (as an example via MVPN Procedures or other mechanism), the following communication happens between the PCE and PCCs¶
PCE on receiving of this report, will compute the P2MP Policy and its replication segments.¶
The following procedures are the same for PCE or PCC Init.¶
Any new candidate path for the P2MP Policy is downloaded by PCE to the Root by sending a PCInitiate message¶
New leaves can be discovered via Multicast procedures, and new replication segments can be instantiated or existing one updated to reach these leaves¶
When a P2MP LSP needs to be optimized for any reason (i.e. it is taking a FRR tunnel or new routers are added to the network) a global optimization of the candidate path is possible.¶
Each Candidate Path can contain two Path-Instances. The current unoptimized Path-Instance is the active instance and its replication segments are forwarding the multicast PDUs from the root to the leaves. However the second optimized Path-Instance will be setup with its own unique replication segments throughout the network, from the Root to the leaves. These two Path-Instances can co-exist. The two Path-Instances are uniquely identified by their Instance-ID in the SR-P2MP-POLICY-ID-TLV (defined in this document). After the optimized LSP has been downloaded successfully PCC MUST send two reports, reporting UP of the new path indicating the new LSP-ID, and a second reporting the tear down of the old path with the R bit of the LSP Object SET with the old Instance-ID in the SR-P2MP-POLICY-ID-TLV. This MBB procedure will move the multicast PDUs to the optimized Path-Instance.¶
The leaf should be able to accept traffic from both Path-Instances to minimize the traffic outage by the Make Before Break process.¶
Currently this draft identifies the Facility FRR procedures. In addition, only LINK Protection procedures are defined. How the Facility Path is built and instantiated is beyond the scope of this document.¶
R | | T | --- | | L1 L2 Figure 1 R---F1 | | T---F2 | --- | | L1 L2 Figure 2¶
As an example, in figure 1 both R and T are configured with replication segments. There are two interface between R and T. One can be used as primary and second as a bypass in case the primary interface is down. There can be 2 method to protect the primary interface.¶
As a second example, in figure 2, R and T connected directly and via network F1..F2. In this example as per example 1 unicast SR can be used to connect the two replication segments and in this case the unicast SR LFA or R-LFA or TI-LFA can be used to protect the direct link between R-T via F1. That said if there is no unicast SR available with in the network, the PCE optionally can setup a shared replication point on F1 and F2 and protect all path-instances that are traversing R-T via this shared replication segment.¶
In addition, PHP procedure and implicit null label on the bypass path can be implemented to reduce the PCE programming on the MP PCC.¶
There could be nodes between two replication segment that do not support P2MP Policy or Replication segment. It is possible to connect two non-adjacent Replication segments via a unicast segment routing path and SID list. The SID list can be comprised of any IGP supported segment types (ex: Binding, Adjacency, Node). This information is encoded via the SR-ERO sub-objects and ERO-attributes objects. The last segment in an encoding SID list MUST be a replication segment¶
SR P2MP Policy can be constructed via the following objects¶
<Common Header>¶
<SRP>¶
<P2MP LSP>¶
[<association-list>]¶
optionally if the root is updating the PCE with end point list the end-point-list object can be added.¶
[<end-points-list>]¶
Replication segment can be constructed via the following objects¶
<Common Header> <SRP> <P2MP LSP> (<cci-list>| (<CCI><intended-path>)) <cci-list> ::= <CCI> [<cci-list>] <intended-path> ::= ((<PATH-ATTRIB><ERO>) [<intended-path>])¶
Path-attribute as per [draft-ietf-pce-multipath]¶
The above new objects and TLV's defined in this document can be included in PCRpt, PcInitiate and PcUpd messages.¶
It should be noted that every PcRpt, PcInitiate and PCUpd messages will contain full list of the Leaves and segment and forwarding information that is needed to build the Candidate path and its Replication segments. They will never send the delta information related to the new leaves or forwarding information that need to be added or updated. This is necessary to ensure that PCE or any new PCE is in sync with the PCC.¶
When a PCRpt, PCInitiate and PCUpd messages is sent via PCEP it maintains the previous ERO Path IDs and generates new Path IDs for new instructions, as per [draft-ietf-pce-multipath]. The PATH IDs are maintained for each specific forwarding instructions until the instructions are deleted. For example: When the first leaf is added, the PCE will update with Path ID 1 to the PCC. When the second leaf is added, according to the path calculated, PCE might just append the existing instruction Path ID 1 with a new Path ID 2 to construct the new PCUpd message.¶
The CCI Object is used to identify the entire cross connect of incoming segment and the set of outgoing Interfaces and their corresponding SIDs/SIDList. Any modification to the cross connect should use this CCI ID to identify the cross connect uniquely. Leaves and their corresponding Path IDs can be removed from the cross connect identified via the CCI. The CC-ID is assigned by the PCE.¶
Format of the open Object:¶
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Ver | Flags | Keepalive | DeadTimer | SID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | // Optional TLVs // | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
All the nodes need to establish a PCEP connection with the PCE.¶
During PCEP Initialization Phase, PCEP Speakers need to set flags N, M, P in the STATEFUL-PCE-CAPABILITY TLV as defined in [draft-ietf-pce-stateful-pce-p2mp] section-5.2¶
This draft extends the PCEP OPEN object by defining an optional TLV to indicate the PCE's capability to perform SR-P2MP path computations with a new IANA capability type.¶
The inclusion of this TLV in an OPEN object indicates that the sender can perform SR-P2MP path computations. This will be similar to the P2MP-CAPABILITY defined in [RFC8306] section-3.1.2 and a new value needs to be defined for SR-P2MP.¶
A Path Setup Type (PST) of PCECC is also added as per [draft-ietf-pce-pcep-extension-for-pce-controller].¶
This document also introduces a new bit S in the SR PCECC capacity Sub TLV indicating the support to handle PCECC based label download for Replication segment.¶
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type=1 | Length=4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Flags |S|M|L| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Also, the N,M,P bits in STATEFUL-PCE-CAPABILITY TLV should be SET.¶
A Assoc-Type-List TLV as per [RFC8697] section 3.4 should be send via PCEP open object with following association type¶
+-------------------+----------------------------+------------------+ | Association Type | Association Name | Reference | | Value | | | +-------------------+----------------------------+------------------+ | TBD1 | P2MP SR Policy Association | This document | +-------------------+----------------------------+------------------+¶
OP-CONF-Assoc-RANGE (Operator-configured Association Range)should not be set for this association type and must be ignored.¶
The open message MUST include the MULTIPATH CAPABILITY TLV as defined in [draft-ietf-pce-multipath]¶
As per [RFC8231] section 7.3.2. a Symbolic Path Name TLV should uniquely identify the P2MP path on the PCC. This symbolic path name is a human-readable string that identifies an P2MP LSP in the network. It needs to be constant through the lifetime of the P2MP path.¶
As an example in the case of P2MP LSP the symbolic name can be p2mp policy name + candidate path name of the LSP.¶
Two ASSOCIATION object types for IPv4 and IPv6 are defined in [RFC8697]. The ASSOCIATION object includes "Association type" indicating the type of the association group. This document adds a new Association type. Association type = TBD1 "P2MP SR Policy Association Type" for SR Policy Association Group (P2MP SRPAG). As per [draft-barth-pce-segment-routing-policy-cp] section 5, three new TLVs are identified to carry association information: P2MP-SRPAG- POL-ID-TLV, P2MP-SRPAG-CPATH-ID-TLV, P2MP-SRPAG-CPATH-ATTR-TLV¶
The P2MP-SRPOLICY-POL-ID TLV is a mandatory TLV for the P2MP-SRPAG Association. Only one P2MP-SRPOLICY-POL-ID TLV can be carried and only the first occurrence is processed and any others MUST be ignored.¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Root | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | TREE-ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Type: TBD2 for "P2MP-SR-POLICY-POL-ID" TLV.¶
Length: 8 or 20, depending on length of End-point (IPv4 or IPv6)¶
Tunnel Sender Address : Can be either IPv4 or IPv6, this value is the value of the root loopback IP.¶
Tree-ID: Tree ID that the replication segment is part of as per draft-ietf-spring-sr-p2mp-policy¶
The P2MP-SRPOLICY-CPATH-ID TLV is a mandatory TLV for the P2MPSRPAG Association. Only one P2MP-SRPOLICY-CPATH-ID TLV can be carried and only the first occurrence is processed and any others MUST be ignored.¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Proto. Origin |Flags |A| Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Originator ASN | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Originator Address | | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Discriminator | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Type: TBD3 for "P2MP-SR-POLICY-CPATH-ID" TLV.¶
Length: 28.¶
Protocol Origin: 8-bit value that encodes the protocol origin, as specified in [I-D.ietf-spring-segment-routing-policy] Section 2.3.¶
Flags : A: This candidate path is active. At any instance only one candidate path can be active. PCC indicates the active candidate path to PCE through this bit. Reserved: MUST be set to zero on transmission and ignored on receipt.¶
Originator ASN: Represented as 4 byte number, part of the originator identifier, as specified in [draft-ietf-spring-segment-routing-policy] Section 2.4.¶
Originator Address: Represented as 128 bit value where IPv4 address are encoded in lowest 32 bits, part of the originator identifier, as specified in [draft-ietf-spring-segment-routing-policy] Section 2.4.¶
Discriminator: 32-bit value that encodes the Discriminator of the candidate path.¶
The P2MP-SRPOLICY-CPATH-ATTR TLV is an optional TLV for the SRPAG Association. Only one P2MP-SRPOLICY-CPATH-ATTR TLV can be carried and only the first occurrence is processed and any others MUST be ignored.¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Preference | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Type: TBD4 for "P2MP-SRPOLICY-CPATH-ATTR" TLV.¶
Length: 4. Preference: Numerical preference of the candidate path, as specified in [draft-ietf-spring-segment-routing-policy] Section 2.7.¶
If the TLV is missing, a default preference of 100 as specified in [draft-ietf-spring-segment-routing-policy] is used.¶
In order for the Root to indicate operations of its leaves(Add/Remove/Modify/DoNotModify), the PC Report message is extended to include P2MP End Point <P2MP End-points> Object which is defined in [RFC8306]¶
The format of the PC Report message is as follow:¶
<Common Header>¶
[<SRP>]¶
<LSP>¶
[<association-list>]¶
[<end-points-list>]¶
IPV4-P2MP END-POINTS: 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Leaf type | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Source IPv4 address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination IPv4 address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ ~ ... ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination IPv4 address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ IPV6-P2MP END-POINTS: 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Leaf type | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Source IPv6 address (16 bytes) | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Destination IPv6 address (16 bytes) | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ ~ ... ~ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Destination IPv6 address (16 bytes) | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Leaf Types (derived from [RFC8306] section 3.3.2) :¶
A given P2MP END-POINTS object gathers the leaves of a given type. Note that a P2MP report can mix the different types of leaves by including several P2MP END-POINTS objects. The END-POINTS object body has a variable length. These are multiples of 4 bytes for IPv4, multiples of 16 bytes, plus 4 bytes, for IPv6.¶
As it was mentioned previously both P2MP Policy and Replication Segment are identified via the LSP object and more precisely via the SR-P2MP-LSPID-TLV¶
The P2MP Policy uses the PLSP-ID to identify the Candidate Paths and the Instance-ID to identify a Path-Instance with in the Candidate path.¶
On other hand the Replication Segment uses the SR-P2MP-LSPID-TLV to identify and correlate a Replication Segment to a P2MP Policy¶
As it was noted previously on the Root, the P2MP Policy and the Replication Segment is downloaded via the same PCUpd message.¶
The LSP Object is defined in Section 7.3 of [RFC8231]. It specifies the PLSP-ID to uniquely identify an LSP that is constant for the life time of a PCEP session. Similarly, for a P2MP tunnel, the PLSP-ID identify a Candidate Path uniquely with in the P2MP policy.¶
The LSP Object MUST include the new SR-P2MP-POLICY-ID-TLV (IPV4/IpV6) defined in this document below. This is a variation to the P2MP object defined in [draft-ietf-pce-stateful-pce-p2mp]¶
SR-IPV4-P2MP-POLICY-ID TLV: 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type=TBD | Length=10 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Root | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Tree-ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Path-Instance-ID | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ SR-IPV6-P2MP-POLICY-ID TLV : 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type=TBD | Length=22 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | + + | Root | + (16 octets) + | | + + | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Tree-ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Path-Instance-ID | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
The type (16-bit) of the TLV is TBD (need allocation by IANA).¶
Root: Source Router IP Address¶
Tree-ID: Unique Identifier of this P2MP LSP on the Root.¶
Instance-ID : Contains 16 Bit instance ID.¶
As per [draft-ietf-spring-sr-replication-segment] a replication segment has a next-hop-group which MAY contain a single outgoing replication SID or a list of SIDs (sr-policy-sid-list) In either case there needs to be a replication SID at the bottom of the stack. This means two replication segments can be directly connected or connected via a SR domain.¶
The format of a Replication Segment message encoding is similar to P2MP Policy. However, the P2MP Policy contains the association object and the replication segment message does not contain the association object. In addition the replication segment uses the CCI object to identify a P2MP cross connect. The replication segment is downloaded individually to the root, transit and leaf nodes without the P2MP Policy. The P2MP Policy is a Root Concept. The replication segment uses SR-P2MP-LSPID-TLV as its identifier. The TLV is coded differently for shared and non-shared case.¶
The CCI Object as defined in [draft-ietf-pce-pcep-extension-for-pce-controller] is used to identify a forwarding instruction in the Replication Segment. A forwarding instruction is incoming SID and a set of outgoing branches. The CCI Object-Type of 1 is used for the MPLS Label. The label in the CCI Object is the incoming SID. The outgoing SIDs are defined by the ERO Objects.¶
The CCI Object can be include in Reports, initiate and Update messages for Replication Segments.¶
The PCInitiate message defined in [RFC8281] and extended in [draft-ietf-pce-pcep-extension-for-pce-controller] is further extended to support SR-P2MP replication segment based central control instructions.¶
The format of the extended PCInitiate message is as follows: <PCInitiate Message> ::= <Common Header> <PCE-initiated-lsp-list> Where: <Common Header> is defined in [RFC5440] <PCE-initiated-lsp-list> ::= <PCE-initiated-lsp-request> [<PCE-initiated-lsp-list>] <PCE-initiated-lsp-request> ::= (<PCE-initiated-lsp-instantiation>| <PCE-initiated-lsp-deletion>| <PCE-initiated-lsp-central-control>) <PCE-initiated-lsp-central-control> ::= <SRP> <LSP> (<cci-list>| (<CCI><intended-path>)) <cci-list> ::= <CCI> [<cci-list>] <intended-path> ::= ((<PATH-ATTRIB><ERO>) [<intended-path>]) Where: <PCE-initiated-lsp-instantiation> and <PCE-initiated-lsp-deletion> are as per [RFC8281].¶
The LSP and SRP object is defined in [RFC8231]. The <intended-path> is as per [RFC8281] [draft-ietf-pce-multipath] (PATH-ATTRIB and ERO).¶
The format of the PCRpt message is as follows: <PCRpt Message> ::= <Common Header> <state-report-list> Where: <state-report-list> ::= <state-report>[<state-report-list>] <state-report> ::= (<lsp-state-report>| <central-control-report>) <lsp-state-report> ::= [<SRP>] <LSP> <path> <central-control-report> ::= [<SRP>] <LSP> (<cci-list>| (<CCI><intended-path>)) <cci-list> ::= <CCI> [<cci-list>] Where: <path> is as per [RFC8231] and the LSP and SRP object are also defined in [RFC8231]. The <intended-path> is as per [draft-ietf-pce-multipath] (PATH-ATTRIB and ERO).¶
This document extends the use of PCUpd message with SR-P2MP CCI as follows:¶
<PCUpd Message> ::= <Common Header> <update-request-list> Where: <update-request-list> ::= <update-request>[<update-request-list>] <update-request> ::= (<lsp-update-request>| <central-control-update>) <lsp-update-request> ::= <SRP> <LSP> <path> <central-control-update> ::= <SRP> <LSP> (<CCI><intended-path>) Where: <path> is as per [RFC8231] and the LSP and SRP object are also defined in [RFC8231]. The <intended-path> is as per [draft-ietf-pce-multipath] (PATH-ATTRIB and ERO).¶
The node action and role of ingress, transit, leaf or bud, is indicated via a new Node Role TLV. This document introduces a new SR-P2MP-NODE-ROLE TLV (Type To be assigned by IANA) that will be present in the PATH-ATTRIB object.¶
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type=TBD | Length=4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Role Type | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Forwarding information of a replication segment can be configured and steered via many different mechanisms.¶
As an example a replication SID can be steered via:¶
Replication SID steered with an IPv4/IPv6 directly connected nexthop¶
Replication SID steered with an IPv4/IPv6 loopback address that reside on the directly connected router.¶
Replication SID steered via a SR adjacency or node SID¶
SR-ERO from RFC 8664 is used to construct the forwarding information needed for Replication Segment.¶
A new D flag was added to indicate a loopback nexthop that is residing on the directly attached router. It should be noted that this flag should be set only for the loopback case and not for a local interface as a nexthop.¶
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |L| Type=36 | Length | NT | Flags |D|F|S|C|M| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | SID (optional) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ // NAI (variable, optional) // +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+¶
Flags : F, S, C, M are already defined in rfc8664.¶
This document defines a new flag D: If the next-hop in NAI field is system IP or loopback, this bit indicates whether the system IP / loopback is directly connected router or not. If set indicates directly connected address. When this bit is set, F bit should be 0 (meaning NAI should be present)¶
To delete the entire tree (P2MP LSP), Root send a PCRpt message with the R bit of the LSP object set and all the fields of the SR-P2MP- LSP-ID TLV set to 0(indicating to remove all state associated with this P2MP tunnel). The PCE in response sends a PCInitiate message with R bit in the SRP object SET to all nodes along the path to indicate deletion of the entries.¶
The Fragmentation bit in the LSP object (F bit) can be used to indicate a fragmented PCEP message¶
PCC-Initiated Workflow¶
+-------+ +-------+ |PCC | | PCE | |Root | +-------+ +------| | | | PCC +-------+ | | Transit| | | +------| | |---PCRpt,PLSP-ID=1,D=1-------------->| PCECC LSP |PCC +--------+ | N=1,root-addr,tree-id=a, | SR-Policy | | | | instance-id =b, | Report to |Leaf | | | p2mp-end-points(LeafType=5) | PCE +--------+ | | | | | | | | | |<--PCUpdate,PLSP-ID=1, SRP-ID =1, | Update | | | root-addr,tree-id=a,instance-id=b,| CP | | | p2mp-end-points, association-obj | | | | | | | |-------PCRpt,PLSP-ID=1, SRP-ID = 1,->| | | | root-addr,tree-id=a,instance-id=b,| | | | p2mp-end-points(LeafType=5) | | | | association-object, | | | | | |<---------------PCInitiate,PLSP-ID=0, -------------| Download | | | root-addr,tree-id=a,instance-id=0,| Leaf | | | CC-ID=Z,C=0, | Replication | | | O=0,L=z,path-attribute,ERO,SR-ERO | Segment(RS) | | | | |---------------------PCRpt,PLSP-ID=1-------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=Z,Label=z,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | |<-------PCInitiate,PLSP-ID=0, -------------| Download | | | root-addr,tree-id=a,instance-id=0,| Transit | | | CC-ID=Y,C=0, | RS | | | O=0,L=y,path-attribute,ERO,SR-ERO | | | | | | |-------------PCRpt,PLSP-ID=2-------------->| | | | root-addr,tree-id=a,instance-id=c,| | | | CC-ID=Y,Label=y,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | | |<--PCInitiate,PLSP-ID=1, | Download | | | root-addr,tree-id=a,instance-id=b,| Root | | | CC-ID=X,C=0, | RS | | | O=0,L=x,p2mp-end- | | | | points(LeafType=5),path-attribute,| | | | ERO,SR-ERO | | | | | | | |-------PCRpt,PLSP-ID=1-------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=X,Label=x,O=0, | | | | p2mp-end-points(LeafType=5) | | | | path-attriute,ERO,SR-ERO | | | | | | | |<--PCUpdate,PLSP-ID=1, SRP-ID =2, | | | | root-addr,tree-id=a,instance-id=b,| Activate | | | p2mp-end-points | CP to last | | | | RS | | |-------PCRpt,PLSP-ID=1, SRP-ID =2, ->| | | | root-addr,tree-id=a,instance-id=b,| | | | p2mp-end-points(LeafType=5) | | | | |¶
Note that on transit / leaf Initiate is with PLSP-ID = 0. Therefore PLSP-ID is locally unique to a node. It should be noted that the CC-ID does not need to be constant across all nodes that make up the path.¶
PCE-Initiated workflow¶
+-------+ +-------+ |PCC | | PCE | |Root | +-------+ +------| | | | PCC +-------+ | | Transit| | | +------| | | | PCECC LSP |PCC +--------+ | | | | | | | |Leaf | | | | +--------+ | | | |<---------------PCInitiate,PLSP-ID=0, -------------| Download | | | root-addr,tree-id=a,instance-id=0,| Leaf RS | | | CC-ID=Z,C=0, | | | | O=0,L=z,path-attribute,ERO,SR-ERO | | | | | |---------------------PCRpt,PLSP-ID=1-------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=Z,Label=z,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | |<-------PCInitiate,PLSP-ID=0, -------------| Download | | | root-addr,tree-id=a,instance-id=0,| Transit RS | | | CC-ID=Y,C=0, | | | | O=0,L=y,path-attribute,ERO,SR-ERO | | | | | | |-------------PCRpt,PLSP-ID=2-------------->| | | | root-addr,tree-id=a,instance-id=c,| | | | CC-ID=Y,Label=y,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | | |<--PCInitiate,PLSP-ID=0, | Initiate | | | root-addr,tree-id=0,instance-id=0,| CP | | | p2mp-end-points, association-obj | | | | | | | |-------PCRpt,PLSP-ID=1,------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | p2mp-end-points(LeafType=5) | | | | association-object, | | | | | | | |<--PCInitiate,PLSP-ID=1, | Download | | | root-addr,tree-id=a,instance-id=b,| Root RS | | | CC-ID=X,C=0, | | | | O=0,L=x,p2mp-end- | | | | points(LeafType=5),path-attribute,| | | | ERO,SR-ERO | | | | | | | |-------PCRpt,PLSP-ID=1-------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=X,Label=x,O=0, | | | | p2mp-end-points(LeafType=5) | | | | path-attriute,ERO,SR-ERO | | | | | | |<-------PCInitiate,PLSP-ID=0, -------------| | | | root-addr,tree-id=a,instance-id=0,| | | | CC-ID=Y,C=0, | | | | O=0,L=y,path-attribute,ERO,SR-ERO | | | | | | |-------------PCRpt,PLSP-ID=2-------------->| | | | root-addr,tree-id=a,instance-id=c,| | | | CC-ID=Y,Label=y,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | | |<--PCUpdate,PLSP-ID=1, SRP-ID =1, | Bind and | | | root-addr,tree-id=a,instance-id=b,| Activate | | | p2mp-end-points, | CP to last | | | | RS | | |-------PCRpt,PLSP-ID=1, SRP-ID = 1,->| | | | root-addr,tree-id=a,instance-id=b,| | | | p2mp-end-points(LeafType=5) |¶
MBB Workflow:¶
Common (PCE-INIT, PCC-INIT) MBB +-------+ +-------+ |PCC | | PCE | |Root | +-------+ +------| | | | PCC +-------+ | | Transit| | | +------| | | | PCECC LSP |PCC +--------+ | | | | | | | |Leaf | | | | +--------+ | | | |<---------------PCInitiate,PLSP-ID=1, -------------| Download | | | root-addr,tree-id=a,instance-id=b,| new RS on | | | CC-ID=Z1,C=0, | Leaf | | | O=0,L=z1,path-attribute,ERO,SR-ERO | | | | | |---------------------PCRpt,PLSP-ID=1-------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=Z1,Label=z1,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | |<-------PCInitiate,PLSP-ID=2, -------------| Download | | | root-addr,tree-id=a,instance-id=c,| new RS on | | | CC-ID=Y1,C=0, | Transit | | | O=0,L=y1,path-attribute,ERO,SR-ERO | | | | | | |-------------PCRpt,PLSP-ID=2-------------->| | | | root-addr,tree-id=a,instance-id=c,| | | | CC-ID=Y1,Label=y1,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | | | |<--PCInitiate,PLSP-ID=1, | Download | | | root-addr,tree-id=a,instance-id=b,| new RS on | | | CC-ID=X1,C=0, | Root | | | O=0,L=x1,p2mp-end- | | | | points(LeafType=5),path-attribute,| | | | ERO,SR-ERO | | | | | | | |-------PCRpt,PLSP-ID=1-------------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=X1,Label=x1,O=0, | | | | p2mp-end-points(LeafType=5) | | | | path-attriute,ERO,SR-ERO | | | | | | | |<--PCUpdate,PLSP-ID=1, SRP-ID =1, | Bind and | | | root-addr,tree-id=a,instance-id=b,| Activate , | | | p2mp-end-points, | CP to last | | | | RS | | |-------PCRpt,PLSP-ID=1, SRP-ID = 1,->| | | | root-addr,tree-id=a,instance-id=b,| | | | p2mp-end-points(LeafType=5) | | | | | | | |<--PCInitiate,PLSP-ID=1,R=1 | Remove | | | root-addr,tree-id=a,instance-id=b,| the old RS | | | CC-ID=X1,C=0 | from Leaf | | | O=0,L=x1,p2mp-end- | | | | points(LeafType=5),path-attribute,| | | | ERO,SR-ERO | | | | | | | |-------PCRpt,PLSP-ID=1, R=1--------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=X1,Label=x1,O=0, | | | | p2mp-end-points(LeafType=5) | | | | path-attriute,ERO,SR-ERO | | | | | | |<-------PCInitiate,PLSP-ID=2, R=1----------| Remove the | | | root-addr,tree-id=a,instance-id=c,| old RS from | | | CC-ID=Y1,C=0, | Transit | | | O=0,L=y1,path-attribute,ERO,SR-ERO | | | | | | |-------------PCRpt,PLSP-ID=2, R=1--------->| | | | root-addr,tree-id=a,instance-id=c,| | | | CC-ID=Y1,Label=y1,O=0, | | | | path-attribute,ERO,SR-ERO | | | | | |<---------------PCInitiate,PLSP-ID=1,R=1-----------| Remove the | | | root-addr,tree-id=a,instance-id=b,| old RS from | | | CC-ID=Z1,C=0, | Root | | | O=0,L=z1,path-attribute,ERO,SR-ERO | | | | | |---------------------PCRpt,PLSP-ID=1,R=1---------->| | | | root-addr,tree-id=a,instance-id=b,| | | | CC-ID=Z1,Label=z1,O=0, | | | | path-attribute,ERO,SR-ERO |¶
A new Association type. Association type = TBD1 "P2MP SR Policy Association Type" for SR Policy Association Group (P2MP SRPAG)¶
TBD¶
The authors would like to thank Tanmoy Kundu and Stone Andrew at Nokia for their feedback and major contribution to this draft.¶