Network Working Group B. Jamoussi, Editor, Nortel Networks Request for Comments: 3212 L. Andersson, Utfors AB Category: Standards Track R. Callon, Juniper Networks R. Dantu, Netrake Corporation L. Wu, Cisco Systems P. Doolan, OTB Consulting Corp. T. Worster N. Feldman, IBM Corp. A. Fredette, ANF Consulting M. Girish, Atoga Systems E. Gray, Sandburst J. Heinanen, Song Networks, Inc. T. Kilty, Newbridge Networks, Inc. A. Malis, Vivace Networks January 2002
Constraint-Based LSP Setup using LDP
Status of this Memo
This document specifies an Internet standards track protocol for the Internet community, and requests discussion and suggestions for improvements. Please refer to the current edition of the "Internet Official Protocol Standards" (STD 1) for the standardization state and status of this protocol. Distribution of this memo is unlimited.
Copyright Notice
Copyright (C) The Internet Society (2002). All Rights Reserved.
Abstract
This document specifies mechanisms and TLVs (Type/Length/Value) for support of CR-LSPs (constraint-based routed Label Switched Path) using LDP (Label Distribution Protocol).
This specification proposes an end-to-end setup mechanism of a CR-LSP initiated by the ingress LSR (Label Switching Router). We also specify mechanisms to provide means for reservation of resources using LDP.
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 [6].
Jamoussi, et al. Standards Track [Page 1]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Table of Contents
1. Introduction....................................................3 2. Constraint-based Routing Overview...............................4 2.1 Strict and Loose Explicit Routes...............................5 2.2 Traffic Characteristics........................................5 2.3 Preemption.....................................................5 2.4 Route Pinning..................................................6 2.5 Resource Class.................................................6 3. Solution Overview...............................................6 3.1 Required Messages and TLVs.....................................7 3.2 Label Request Message..........................................7 3.3 Label Mapping Message..........................................9 3.4 Notification Message..........................................10 3.5 Release , Withdraw, and Abort Messages........................11 4. Protocol Specification.........................................11 4.1 Explicit Route TLV (ER-TLV)...................................11 4.2 Explicit Route Hop TLV (ER-Hop TLV)...........................12 4.3 Traffic Parameters TLV........................................13 4.3.1 Semantics...................................................15 4.3.1.1 Frequency.................................................15 4.3.1.2 Peak Rate.................................................16 4.3.1.3 Committed Rate............................................16 4.3.1.4 Excess Burst Size.........................................16 4.3.1.5 Peak Rate Token Bucket....................................16 4.3.1.6 Committed Data Rate Token Bucket..........................17 4.3.1.7 Weight....................................................18 4.3.2 Procedures..................................................18 4.3.2.1 Label Request Message.....................................18 4.3.2.2 Label Mapping Message.....................................18 4.3.2.3 Notification Message......................................19 4.4 Preemption TLV................................................19 4.5 LSPID TLV.....................................................20 4.6 Resource Class (Color) TLV....................................21 4.7 ER-Hop semantics..............................................22 4.7.1. ER-Hop 1: The IPv4 prefix..................................22 4.7.2. ER-Hop 2: The IPv6 address.................................23 4.7.3. ER-Hop 3: The autonomous system number....................24 4.7.4. ER-Hop 4: LSPID............................................24 4.8. Processing of the Explicit Route TLV.........................26 4.8.1. Selection of the next hop..................................26 4.8.2. Adding ER-Hops to the explicit route TLV...................27 4.9 Route Pinning TLV.............................................28 4.10 CR-LSP FEC Element...........................................28 5. IANA Considerations............................................29 5.1 TLV Type Name Space...........................................29 5.2 FEC Type Name Space...........................................30 5.3 Status Code Space.............................................30
Jamoussi, et al. Standards Track [Page 2]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
6. Security Considerations........................................31 7. Acknowledgments................................................31 8. Intellectual Property Consideration............................31 9. References.....................................................32 Appendix A: CR-LSP Establishment Examples.........................33 A.1 Strict Explicit Route Example.................................33 A.2 Node Groups and Specific Nodes Example........................34 Appendix B. QoS Service Examples..................................36 B.1 Service Examples..............................................36 B.2 Establishing CR-LSP Supporting Real-Time Applications.........38 B.3 Establishing CR-LSP Supporting Delay Insensitive Applications.38 Author's Addresses................................................39 Full Copyright Statement..........................................42
Label Distribution Protocol (LDP) is defined in [1] for distribution of labels inside one MPLS domain. One of the most important services that may be offered using MPLS in general and LDP in particular is support for constraint-based routing of traffic across the routed network. Constraint-based routing offers the opportunity to extend the information used to setup paths beyond what is available for the routing protocol. For instance, an LSP can be setup based on explicit route constraints, QoS constraints, and other constraints. Constraint-based routing (CR) is a mechanism used to meet Traffic Engineering requirements that have been proposed by, [2] and [3]. These requirements may be met by extending LDP for support of constraint-based routed label switched paths (CR-LSPs). Other uses for CR-LSPs include MPLS-based VPNs [4]. More information about the applicability of CR-LDP can be found in [5].
The need for constraint-based routing (CR) in MPLS has been explored elsewhere [2], and [3]. Explicit routing is a subset of the more general constraint-based routing function. At the MPLS WG meeting held during the Washington IETF (December 1997) there was consensus that LDP should support explicit routing of LSPs with provision for indication of associated (forwarding) priority. In the Chicago meeting (August 1998), a decision was made that support for explicit path setup in LDP will be moved to a separate document. This document provides that support and it has been accepted as a working document in the Orlando meeting (December 1998).
Jamoussi, et al. Standards Track [Page 3]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
This specification proposes an end-to-end setup mechanism of a constraint-based routed LSP (CR-LSP) initiated by the ingress LSR. We also specify mechanisms to provide means for reservation of resources using LDP.
This document introduce TLVs and procedures that provide support for:
- Strict and Loose Explicit Routing - Specification of Traffic Parameters - Route Pinning - CR-LSP Preemption though setup/holding priorities - Handling Failures - LSPID - Resource Class
Section 2 introduces the various constraints defined in this specification. Section 3 outlines the CR-LDP solution. Section 4 defines the TLVs and procedures used to setup constraint-based routed label switched paths. Appendix A provides several examples of CR-LSP path setup. Appendix B provides Service Definition Examples.
Constraint-based routing is a mechanism that supports the Traffic Engineering requirements defined in [3]. Explicit Routing is a subset of the more general constraint-based routing where the constraint is the explicit route (ER). Other constraints are defined to provide a network operator with control over the path taken by an LSP. This section is an overview of the various constraints supported by this specification.
Like any other LSP a CR-LSP is a path through an MPLS network. The difference is that while other paths are setup solely based on information in routing tables or from a management system, the constraint-based route is calculated at one point at the edge of network based on criteria, including but not limited to routing information. The intention is that this functionality shall give desired special characteristics to the LSP in order to better support the traffic sent over the LSP. The reason for setting up CR-LSPs might be that one wants to assign certain bandwidth or other Service Class characteristics to the LSP, or that one wants to make sure that alternative routes use physically separate paths through the network.
Jamoussi, et al. Standards Track [Page 4]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
An explicit route is represented in a Label Request Message as a list of nodes or groups of nodes along the constraint-based route. When the CR-LSP is established, all or a subset of the nodes in a group may be traversed by the LSP. Certain operations to be performed along the path can also be encoded in the constraint-based route.
The capability to specify, in addition to specified nodes, groups of nodes, of which a subset will be traversed by the CR-LSP, allows the system a significant amount of local flexibility in fulfilling a request for a constraint-based route. This allows the generator of the constraint-based route to have some degree of imperfect information about the details of the path.
The constraint-based route is encoded as a series of ER-Hops contained in a constraint-based route TLV. Each ER-Hop may identify a group of nodes in the constraint-based route. A constraint-based route is then a path including all of the identified groups of nodes in the order in which they appear in the TLV.
To simplify the discussion, we call each group of nodes an "abstract node". Thus, we can also say that a constraint-based route is a path including all of the abstract nodes, with the specified operations occurring along that path.
The traffic characteristics of a path are described in the Traffic Parameters TLV in terms of a peak rate, committed rate, and service granularity. The peak and committed rates describe the bandwidth constraints of a path while the service granularity can be used to specify a constraint on the delay variation that the CR-LDP MPLS domain may introduce to a path's traffic.
CR-LDP signals the resources required by a path on each hop of the route. If a route with sufficient resources can not be found, existing paths may be rerouted to reallocate resources to the new path. This is the process of path preemption. Setup and holding priorities are used to rank existing paths (holding priority) and the new path (setup priority) to determine if the new path can preempt an existing path.
The setupPriority of a new CR-LSP and the holdingPriority attributes of the existing CR-LSP are used to specify priorities. Signaling a higher holding priority express that the path, once it has been
Jamoussi, et al. Standards Track [Page 5]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
established, should have a lower chance of being preempted. Signaling a higher setup priority expresses the expectation that, in the case that resource are unavailable, the path is more likely to preempt other paths. The exact rules determining bumping are an aspect of network policy.
The allocation of setup and holding priority values to paths is an aspect of network policy.
The setup and holding priority values range from zero (0) to seven (7). The value zero (0) is the priority assigned to the most important path. It is referred to as the highest priority. Seven (7) is the priority for the least important path. The use of default priority values is an aspect of network policy. The recommended default value is (4).
The setupPriority of a CR-LSP should not be higher (numerically less) than its holdingPriority since it might bump an LSP and be bumped by the next "equivalent" request.
Route pinning is applicable to segments of an LSP that are loosely routed - i.e. those segments which are specified with a next hop with the "L" bit set or where the next hop is an abstract node. A CR-LSP may be setup using route pinning if it is undesirable to change the path used by an LSP even when a better next hop becomes available at some LSR along the loosely routed portion of the LSP.
The network operator may classify network resources in various ways. These classes are also known as "colors" or "administrative groups". When a CR-LSP is being established, it's necessary to indicate which resource classes the CR-LSP can draw from.
CR-LSP over LDP Specification is designed with the following goals:
1. Meet the requirements outlined in [3] for performing traffic engineering and provide a solid foundation for performing more general constraint-based routing.
2. Build on already specified functionality that meets the requirements whenever possible. Hence, this specification is based on [1].
Jamoussi, et al. Standards Track [Page 6]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
In this document, support for unidirectional point-to-point CR-LSPs is specified. Support for point-to-multipoint, multipoint-to-point, is for further study (FFS).
Support for constraint-based routed LSPs in this specification depends on the following minimal LDP behaviors as specified in [1]:
- Use of Basic and/or Extended Discovery Mechanisms. - Use of the Label Request Message defined in [1] in downstream on demand label advertisement mode with ordered control. - Use of the Label Mapping Message defined in [1] in downstream on demand mode with ordered control. - Use of the Notification Message defined in [1]. - Use of the Withdraw and Release Messages defined in [1]. - Use of the Loop Detection (in the case of loosely routed segments of a CR-LSP) mechanisms defined in [1].
In addition, the following functionality is added to what's defined in [1]:
- The Label Request Message used to setup a CR-LSP includes one or more CR-TLVs defined in Section 4. For instance, the Label Request Message may include the ER-TLV.
- An LSR implicitly infers ordered control from the existence of one or more CR-TLVs in the Label Request Message. This means that the LSR can still be configured for independent control for LSPs established as a result of dynamic routing. However, when a Label Request Message includes one or more of the CR- TLVs, then ordered control is used to setup the CR-LSP. Note that this is also true for the loosely routed parts of a CR- LSP.
- New status codes are defined to handle error notification for failure of established paths specified in the CR-TLVs. All of the new status codes require that the F bit be set.
Optional TLVs MUST be implemented to be compliant with the protocol. However, they are optionally carried in the CR-LDP messages to signal certain characteristics of the CR-LSP being established or modified.
Examples of CR-LSP establishment are given in Appendix A to illustrate how the mechanisms described in this document work.
Jamoussi, et al. Standards Track [Page 7]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Any Messages, TLVs, and procedures not defined explicitly in this document are defined in the LDP Specification [1]. The reader can use [7] as an informational document about the state transitions, which relate to CR-LDP messages.
The following subsections are meant as a cross-reference to the [1] document and indication of additional functionality beyond what's defined in [1] where necessary.
Note that use of the Status TLV is not limited to Notification messages as specified in Section 3.4.6 of [1]. A message other than a Notification message may carry a Status TLV as an Optional Parameter. When a message other than a Notification carries a Status TLV the U-bit of the Status TLV should be set to 1 to indicate that the receiver should silently discard the TLV if unprepared to handle it.
The Label Request Message is as defined in 3.5.8 of [1] with the following modifications (required only if any of the CR-TLVs is included in the Label Request Message):
- The Label Request Message MUST include a single FEC-TLV element. The CR-LSP FEC TLV element SHOULD be used. However, the other FEC- TLVs defined in [1] MAY be used instead for certain applications.
- The Optional Parameters TLV includes the definition of any of the Constraint-based TLVs specified in Section 4.
- The Procedures to handle the Label Request Message are augmented by the procedures for processing of the CR-TLVs as defined in Section 4.
Jamoussi, et al. Standards Track [Page 8]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The encoding for the CR-LDP Label Request Message is as follows:
The Notification Message is as defined in Section 3.5.1 of [1] and the Status TLV encoding is as defined in Section 3.4.6 of [1]. Establishment of an CR-LSP may fail for a variety of reasons. All such failures are considered advisory conditions and they are signaled by the Notification Message.
Notification Messages carry Status TLVs to specify events being signaled. New status codes are defined in Section 4.11 to signal error notifications associated with the establishment of a CR-LSP and the processing of the CR-TLV. All of the new status codes require that the F bit be set.
The Notification Message MAY carry the LSPID TLV of the corresponding CR-LSP.
Notification Messages MUST be forwarded toward the LSR originating the Label Request at each hop and at any time that procedures in this specification - or in [1] - specify sending of a Notification Message in response to a Label Request Message.
Jamoussi, et al. Standards Track [Page 10]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The encoding of the notification message is as follows:
The Label Request Message defined in [1] MUST carry the LSPID TLV and MAY carry one or more of the optional Constraint-based Routing TLVs (CR-TLVs) defined in this section. If needed, other constraints can be supported later through the definition of new TLVs. In this specification, the following TLVs are defined:
The ER-TLV is an object that specifies the path to be taken by the LSP being established. It is composed of one or more Explicit Route Hop TLVs (ER-Hop TLVs) defined in Section 4.2.
Jamoussi, et al. Standards Track [Page 11]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The contents of an ER-TLV are a series of variable length ER-Hop TLVs.
A node receiving a label request message including an ER-Hop type that is not supported MUST not progress the label request message to the downstream LSR and MUST send back a "No Route" Notification Message.
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
ER-Hop Type A fourteen-bit field carrying the type of the ER-Hop contents. Currently defined values are:
Value Type ------ ------------------------ 0x0801 IPv4 prefix 0x0802 IPv6 prefix 0x0803 Autonomous system number 0x0804 LSPID
Length Specifies the length of the value field in bytes.
L bit The L bit in the ER-Hop is a one-bit attribute. If the L bit is set, then the value of the attribute is "loose." Otherwise, the value of the attribute is "strict." For brevity, we say that if the value of the ER-Hop attribute is loose then it is a "loose ER-Hop." Otherwise, it's a "strict ER-Hop." Further, we say that the abstract node of a strict or loose ER-Hop is a strict or a loose node, respectively. Loose and strict nodes are always interpreted relative to their prior abstract nodes. The path between a strict node and its prior node MUST include only network nodes from the strict node and its prior abstract node.
The path between a loose node and its prior node MAY include other network nodes, which are not part of the strict node or its prior abstract node.
Contents A variable length field containing a node or abstract node which is one of the consecutive nodes that make up the explicitly routed LSP.
Type A fourteen-bit field carrying the value of the Traffic Parameters TLV Type = 0x0810.
Length Specifies the length of the value field in bytes = 24.
Flags The Flags field is shown below:
+--+--+--+--+--+--+--+--+ | Res |F6|F5|F4|F3|F2|F1| +--+--+--+--+--+--+--+--+
Res - These bits are reserved. Zero on transmission. Ignored on receipt. F1 - Corresponds to the PDR. F2 - Corresponds to the PBS. F3 - Corresponds to the CDR. F4 - Corresponds to the CBS. F5 - Corresponds to the EBS. F6 - Corresponds to the Weight.
Jamoussi, et al. Standards Track [Page 14]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Each flag Fi is a Negotiable Flag corresponding to a Traffic Parameter. The Negotiable Flag value zero denotes NotNegotiable and value one denotes Negotiable.
Frequency The Frequency field is coded as an 8 bit unsigned integer with the following code points defined:
0- Unspecified 1- Frequent 2- VeryFrequent 3-255 - Reserved Reserved - Zero on transmission. Ignored on receipt.
Weight An 8 bit unsigned integer indicating the weight of the CR-LSP. Valid weight values are from 1 to 255. The value 0 means that weight is not applicable for the CR-LSP.
Traffic Parameters Each Traffic Parameter is encoded as a 32-bit IEEE single- precision floating-point number. A value of positive infinity is represented as an IEEE single-precision floating-point number with an exponent of all ones (255) and a sign and mantissa of all zeros. The values PDR and CDR are in units of bytes per second. The values PBS, CBS and EBS are in units of bytes.
The value of PDR MUST be greater than or equal to the value of CDR in a correctly encoded Traffic Parameters TLV.
The Frequency specifies at what granularity the CDR allocated to the CR-LSP is made available. The value VeryFrequent means that the available rate should average at least the CDR when measured over any time interval equal to or longer than the shortest packet time at the CDR. The value Frequent means that the available rate should average at least the CDR when measured over any time interval equal to or longer than a small number of shortest packet times at the CDR.
The value Unspecified means that the CDR MAY be provided at any granularity.
Jamoussi, et al. Standards Track [Page 15]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The Peak Rate defines the maximum rate at which traffic SHOULD be sent to the CR-LSP. The Peak Rate is useful for the purpose of resource allocation. If resource allocation within the MPLS domain depends on the Peak Rate value then it should be enforced at the ingress to the MPLS domain.
The Peak Rate is defined in terms of the two Traffic Parameters PDR and PBS, see section 4.3.1.5 below.
The Excess Burst Size may be used at the edge of an MPLS domain for the purpose of traffic conditioning. The EBS MAY be used to measure the extent by which the traffic sent on a CR-LSP exceeds the committed rate.
The possible traffic conditioning actions, such as passing, marking or dropping, are specific to the MPLS domain.
The Excess Burst Size is defined together with the Committed Rate, see section 4.3.1.6 below.
The Peak Rate of a CR-LSP is specified in terms of a token bucket P with token rate PDR and maximum token bucket size PBS.
The token bucket P is initially (at time 0) full, i.e., the token count Tp(0) = PBS. Thereafter, the token count Tp, if less than PBS, is incremented by one PDR times per second. When a packet of size B bytes arrives at time t, the following happens:
- If Tp(t)-B >= 0, the packet is not in excess of the peak rate and Tp is decremented by B down to the minimum value of 0, else
- the packet is in excess of the peak rate and Tp is not decremented.
Jamoussi, et al. Standards Track [Page 16]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Note that according to the above definition, a positive infinite value of either PDR or PBS implies that arriving packets are never in excess of the peak rate.
The actual implementation of an LSR doesn't need to be modeled according to the above formal token bucket specification.
The committed rate of a CR-LSP is specified in terms of a token bucket C with rate CDR. The extent by which the offered rate exceeds the committed rate MAY be measured in terms of another token bucket E, which also operates at rate CDR. The maximum size of the token bucket C is CBS and the maximum size of the token bucket E is EBS.
The token buckets C and E are initially (at time 0) full, i.e., the token count Tc(0) = CBS and the token count Te(0) = EBS.
Thereafter, the token counts Tc and Te are updated CDR times per second as follows:
- If Tc is less than CBS, Tc is incremented by one, else - if Te is less then EBS, Te is incremented by one, else neither Tc nor Te is incremented.
When a packet of size B bytes arrives at time t, the following happens:
- If Tc(t)-B >= 0, the packet is not in excess of the Committed Rate and Tc is decremented by B down to the minimum value of 0, else
- if Te(t)-B >= 0, the packet is in excess of the Committed rate but is not in excess of the EBS and Te is decremented by B down to the minimum value of 0, else
- the packet is in excess of both the Committed Rate and the EBS and neither Tc nor Te is decremented.
Note that according to the above specification, a CDR value of positive infinity implies that arriving packets are never in excess of either the Committed Rate or EBS. A positive infinite value of either CBS or EBS implies that the respective limit cannot be exceeded.
The actual implementation of an LSR doesn't need to be modeled according to the above formal specification.
Jamoussi, et al. Standards Track [Page 17]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The weight determines the CR-LSP's relative share of the possible excess bandwidth above its committed rate. The definition of "relative share" is MPLS domain specific.
If an LSR receives an incorrectly encoded Traffic Parameters TLV in which the value of PDR is less than the value of CDR then it MUST send a Notification Message including the Status code "Traffic Parameters Unavailable" to the upstream LSR from which it received the erroneous message.
If a Traffic Parameter is indicated as Negotiable in the Label Request Message by the corresponding Negotiable Flag then an LSR MAY replace the Traffic Parameter value with a smaller value.
If the Weight is indicated as Negotiable in the Label Request Message by the corresponding Negotiable Flag then an LSR may replace the Weight value with a lower value (down to 0).
If, after possible Traffic Parameter negotiation, an LSR can support the CR-LSP Traffic Parameters then the LSR MUST reserve the corresponding resources for the CR-LSP.
If, after possible Traffic Parameter negotiation, an LSR cannot support the CR-LSP Traffic Parameters then the LSR MUST send a Notification Message that contains the "Resource Unavailable" status code.
If an LSR receives an incorrectly encoded Traffic Parameters TLV in which the value of PDR is less than the value of CDR then it MUST send a Label Release message containing the Status code "Traffic Parameters Unavailable" to the LSR from which it received the erroneous message. In addition, the LSP should send a Notification Message upstream with the status code 'Label Request Aborted'.
If the negotiation flag was set in the label request message, the egress LSR MUST include the (possibly negotiated) Traffic Parameters and Weight in the Label Mapping message.
The Traffic Parameters and the Weight in a Label Mapping message MUST be forwarded unchanged.
Jamoussi, et al. Standards Track [Page 18]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
An LSR SHOULD adjust the resources that it reserved for a CR-LSP when it receives a Label Mapping Message if the Traffic Parameters differ from those in the corresponding Label Request Message.
If an LSR receives a Notification Message for a CR-LSP, it SHOULD release any resources that it possibly had reserved for the CR-LSP. In addition, on receiving a Notification Message from a Downstream LSR that is associated with a Label Request from an upstream LSR, the local LSR MUST propagate the Notification message using the procedures in [1]. Further the F bit MUST be set.
The default value of the setup and holding priorities should be in the middle of the range (e.g., 4) so that this feature can be turned on gradually in an operational network by increasing or decreasing the priority starting at the middle of the range.
Since the Preemption TLV is an optional TLV, LSPs that are setup without an explicitly signaled preemption TLV SHOULD be treated as LSPs with the default setup and holding priorities (e.g., 4).
When an established LSP is preempted, the LSR that initiates the preemption sends a Withdraw Message upstream and a Release Message downstream.
When an LSP in the process of being established (outstanding Label Request without getting a Label Mapping back) is preempted, the LSR that initiates the preemption, sends a Notification Message upstream and an Abort Message downstream.
Type A fourteen-bit field carrying the value of the Preemption-TLV Type = 0x0820.
Length Specifies the length of the value field in bytes = 4.
Jamoussi, et al. Standards Track [Page 19]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Reserved Zero on transmission. Ignored on receipt.
SetPrio A SetupPriority of value zero (0) is the priority assigned to the most important path. It is referred to as the highest priority. Seven (7) is the priority for the least important path. The higher the setup priority, the more paths CR-LDP can bump to set up the path. The default value should be 4.
HoldPrio A HoldingPriority of value zero (0) is the priority assigned to the most important path. It is referred to as the highest priority. Seven (7) is the priority for the least important path. The default value should be 4. The higher the holding priority, the less likely it is for CR- LDP to reallocate its bandwidth to a new path.
LSPID is a unique identifier of a CR-LSP within an MPLS network.
The LSPID is composed of the ingress LSR Router ID (or any of its own Ipv4 addresses) and a Locally unique CR-LSP ID to that LSR.
The LSPID is useful in network management, in CR-LSP repair, and in using an already established CR-LSP as a hop in an ER-TLV.
An "action indicator flag" is carried in the LSPID TLV. This "action indicator flag" indicates explicitly the action that should be taken if the LSP already exists on the LSR receiving the message.
After a CR-LSP is set up, its bandwidth reservation may need to be changed by the network operator, due to the new requirements for the traffic carried on that CR-LSP. The "action indicator flag" is used indicate the need to modify the bandwidth and possibly other parameters of an established CR-LSP without service interruption. This feature has application in dynamic network resources management where traffic of different priorities and service classes is involved.
The procedure for the code point "modify" is defined in [8]. The procedures for other flags are FFS.
Jamoussi, et al. Standards Track [Page 20]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Type A fourteen-bit field carrying the value of the LSPID-TLV Type = 0x0821.
Length Specifies the length of the value field in bytes = 4.
ActFlg Action Indicator Flag: A 4-bit field that indicates explicitly the action that should be taken if the LSP already exists on the LSR receiving the message. A set of indicator code points is proposed as follows:
The Resource Class as defined in [3] is used to specify which links are acceptable by this CR-LSP. This information allows for the network's topology to be pruned.
Jamoussi, et al. Standards Track [Page 21]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The abstract node represented by this ER-Hop is the set of nodes, which have an IP address, which lies within this prefix. Note that a prefix length of 32 indicates a single IPv4 node.
The LSPID is used to identify the tunnel ingress point as the next hop in the ER. This ER-Hop allows for stacking new CR-LSPs within an already established CR-LSP. It also allows for splicing the CR-LSP being established with an existing CR-LSP.
If an LSPID Hop is the last ER-Hop in an ER-TLV, than the LSR may splice the CR-LSP of the incoming Label Request to the CR-LSP that currently exists with this LSPID. This is useful, for example, at the point at which a Label Request used for local repair arrives at the next ER-Hop after the loosely specified CR-LSP segment. Use of the LSPID Hop in this scenario eliminates the need for ER-Hops to keep the entire remaining ER-TLV at each LSR that is at either (upstream or downstream) end of a loosely specified CR-LSP segment as part of its state information. This is due to the fact that the
Jamoussi, et al. Standards Track [Page 24]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
upstream LSR needs only to keep the next ER-Hop and the LSPID and the downstream LSR needs only to keep the LSPID in order for each end to be able to recognize that the same LSP is being identified.
If the LSPID Hop is not the last hop in an ER-TLV, the LSR must remove the LSP-ID Hop and forward the remaining ER-TLV in a Label Request message using an LDP session established with the LSR that is the specified CR-LSP's egress. That LSR will continue processing of the CR-LSP Label Request Message. The result is a tunneled, or stacked, CR-LSP.
To support labels negotiated for tunneled CR-LSP segments, an LDP session is required [1] between tunnel end points - possibly using the existing CR-LSP. Use of the existence of the CR-LSP in lieu of a session, or other possible session-less approaches, is FFS.
A Label Request Message containing an explicit route TLV must determine the next hop for this path. Selection of this next hop may involve a selection from a set of possible alternatives. The mechanism for making a selection from this set is implementation dependent and is outside of the scope of this specification. Selection of particular paths is also outside of the scope of this specification, but it is assumed that each node will make a best effort attempt to determine a loop-free path. Note that such best efforts may be overridden by local policy.
To determine the next hop for the path, a node performs the following steps:
1. The node receiving the Label Request Message must first evaluate the first ER-Hop. If the L bit is not set in the first ER-Hop and if the node is not part of the abstract node described by the first ER-Hop, it has received the message in error, and should return a "Bad Initial ER-Hop Error" status. If the L bit is set and the local node is not part of the abstract node described by the first ER-Hop, the node selects a next hop that is along the path to the abstract node described by the first ER-Hop. If there is no first ER-Hop, the message is also in error and the system should return a "Bad Explicit Routing TLV Error" status using a Notification Message sent upstream.
2. If there is no second ER-Hop, this indicates the end of the explicit route. The explicit route TLV should be removed from the Label Request Message. This node may or may not be the end of the LSP. Processing continues with section 4.8.2, where a new explicit route TLV may be added to the Label Request Message.
3. If the node is also a part of the abstract node described by the second ER-Hop, then the node deletes the first ER-Hop and continues processing with step 2, above. Note that this makes the second ER-Hop into the first ER-Hop of the next iteration.
4. The node determines if it is topologically adjacent to the abstract node described by the second ER-Hop. If so, the node selects a particular next hop which is a member of the abstract node. The node then deletes the first ER-Hop and continues processing with section 4.8.2.
Jamoussi, et al. Standards Track [Page 26]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
5. Next, the node selects a next hop within the abstract node of the first ER-Hop that is along the path to the abstract node of the second ER-Hop. If no such path exists then there are two cases:
5.a If the second ER-Hop is a strict ER-Hop, then there is an error and the node should return a "Bad Strict Node Error" status.
5.b Otherwise, if the second ER-Hop is a loose ER-Hop, then the node selects any next hop that is along the path to the next abstract node. If no path exists within the MPLS domain, then there is an error, and the node should return a "Bad Loose Node Error" status.
6. Finally, the node replaces the first ER-Hop with any ER-Hop that denotes an abstract node containing the next hop. This is necessary so that when the explicit route is received by the next hop, it will be accepted.
7. Progress the Label Request Message to the next hop.
After selecting a next hop, the node may alter the explicit route in the following ways.
If, as part of executing the algorithm in section 4.8.1, the explicit route TLV is removed, the node may add a new explicit route TLV.
Otherwise, if the node is a member of the abstract node for the first ER-Hop, then a series of ER-Hops may be inserted before the first ER-Hop or may replace the first ER-Hop. Each ER-Hop in this series must denote an abstract node that is a subset of the current abstract node.
Alternately, if the first ER-Hop is a loose ER-Hop, an arbitrary series of ER-Hops may be inserted prior to the first ER-Hop.
Jamoussi, et al. Standards Track [Page 27]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
A new FEC element is introduced in this specification to support CR- LSPs. A FEC TLV containing a FEC of Element type CR-LSP (0x04) is a CR-LSP FEC TLV. The CR-LSP FEC Element is an opaque FEC to be used only in Messages of CR-LSPs.
A single FEC element MUST be included in the Label Request Message. The FEC Element SHOULD be the CR-LSP FEC Element. However, one of the other FEC elements (Type=0x01, 0x02, 0x03) defined in [1] MAY be in CR-LDP messages instead of the CR-LSP FEC Element for certain applications. A FEC TLV containing a FEC of Element type CR-LSP (0x04) is a CR-LSP FEC TLV.
FEC Element Type Value Type name
CR-LSP 0x04 No value; i.e., 0 value octets;
Jamoussi, et al. Standards Track [Page 28]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
RFC 3036 [1] defines the LDP TLV name space. This document further subdivides the range of RFC 3036 from that TLV space for TLVs associated with the CR-LDP in the range 0x0800 - 0x08FF.
Following the policies outlined in [IANA], TLV types in this range are allocated through an IETF Consensus action.
Jamoussi, et al. Standards Track [Page 29]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Initial values for this range are specified in the following table:
RFC 3036 defines the FEC Type name space. Further, RFC 3036 has assigned values 0x00 through 0x03. FEC types 0 through 127 are available for assignment through IETF consensus action. This specification makes the following additional assignment, using the policies outlined in [IANA]:
FEC Element Type -------------------------------------- ---------- CR-LSP FEC Element 0x04
RFC 3036 defines the Status Code name space. This document further subdivides the range of RFC 3036 from that TLV space for TLVs associated with the CR-LDP in the range 0x04000000 - 0x040000FF.
Following the policies outlined in [IANA], TLV types in this range are allocated through an IETF Consensus action.
Jamoussi, et al. Standards Track [Page 30]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Initial values for this range are specified in the following table:
Status Code Type -------------------------------------- ----------
Bad Explicit Routing TLV Error 0x04000001 Bad Strict Node Error 0x04000002 Bad Loose Node Error 0x04000003 Bad Initial ER-Hop Error 0x04000004 Resource Unavailable 0x04000005 Traffic Parameters Unavailable 0x04000006 LSP Preempted 0x04000007 Modify Request Not Supported 0x04000008
CR-LDP inherits the same security mechanism described in Section 4.0 of [1] to protect against the introduction of spoofed TCP segments into LDP session connection streams.
7. Acknowledgments
The messages used to signal the CR-LSP setup are based on the work done by the LDP [1] design team.
The list of authors provided with this document is a reduction of the original list. Currently listed authors wish to acknowledge that a substantial amount was also contributed to this work by:
Osama Aboul-Magd, Peter Ashwood-Smith, Joel Halpern, Fiffi Hellstrand, Kenneth Sundell and Pasi Vaananen.
The authors would also like to acknowledge the careful review and comments of Ken Hayward, Greg Wright, Geetha Brown, Brian Williams, Paul Beaubien, Matthew Yuen, Liam Casey, Ankur Anand and Adrian Farrel.
The IETF has been notified of intellectual property rights claimed in regard to some or all of the specification contained in this document. For more information consult the online list of claimed rights.
Jamoussi, et al. Standards Track [Page 31]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
This appendix provides an example for the setup of a strictly routed CR-LSP. In this example, a specific node represents each abstract node.
The sample network used here is a four node network with two edge LSRs and two core LSRs as follows:
abc LSR1------LSR2------LSR3------LSR4
LSR1 generates a Label Request Message as described in Section 3.1 of this document and sends it to LSR2. This message includes the CR- TLV.
A vector of three ER-Hop TLVs <a, b, c> composes the ER-TLV. The ER- Hop TLVs used in this example are of type 0x0801 (IPv4 prefix) with a prefix length of 32. Hence, each ER-Hop TLV identifies a specific node as opposed to a group of nodes. At LSR2, the following processing of the ER-TLV per Section 4.8.1 of this document takes place:
1. The node LSR2 is part of the abstract node described by the first hop <a>. Therefore, the first step passes the test. Go to step 2.
2. There is a second ER-Hop, <b>. Go to step 3.
3. LSR2 is not part of the abstract node described by the second ER-Hop <b>. Go to Step 4.
4. LSR2 determines that it is topologically adjacent to the abstract node described by the second ER-Hop <b>. LSR2 selects a next hop (LSR3) which is the abstract node. LSR2 deletes the first ER-Hop <a> from the ER-TLV, which now becomes <b, c>. Processing continues with Section 4.8.2.
At LSR2, the following processing of Section 4.8.2 takes place: Executing algorithm 4.8.1 did not result in the removal of the ER- TLV.
Also, LSR2 is not a member of the abstract node described by the first ER-Hop <b>.
Finally, the first ER-Hop <b> is a strict hop.
Jamoussi, et al. Standards Track [Page 33]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Therefore, processing section 4.8.2 does not result in the insertion of new ER-Hops. The selection of the next hop has been already done is step 4 of Section 4.8.1 and the processing of the ER-TLV is completed at LSR2. In this case, the Label Request Message including the ER-TLV <b, c> is progressed by LSR2 to LSR3.
At LSR3, a similar processing to the ER-TLV takes place except that the incoming ER-TLV = <b, c> and the outgoing ER-TLV is <c>.
At LSR4, the following processing of section 4.8.1 takes place:
1. The node LSR4 is part of the abstract node described by the first hop <c>. Therefore, the first step passes the test. Go to step 2.
2. There is no second ER-Hop, this indicates the end of the CR- LSP. The ER-TLV is removed from the Label Request Message. Processing continues with Section 4.8.2.
At LSR4, the following processing of Section 4.8.2 takes place: Executing algorithm 4.8.1 resulted in the removal of the ER-TLV. LSR4 does not add a new ER-TLV.
Therefore, processing section 4.8.2 does not result in the insertion of new ER-Hops. This indicates the end of the CR-LSP and the processing of the ER-TLV is completed at LSR4.
At LSR4, processing of Section 3.2 is invoked. The first condition is satisfied (LSR4 is the egress end of the CR-LSP and upstream mapping has been requested). Therefore, a Label Mapping Message is generated by LSR4 and sent to LSR3.
At LSR3, the processing of Section 3.2 is invoked. The second condition is satisfied (LSR3 received a mapping from its downstream next hop LSR4 for a CR-LSP for which an upstream request is still pending). Therefore, a Label Mapping Message is generated by LSR3 and sent to LSR2.
At LSR2, a similar processing to LSR 3 takes place and a Label Mapping Message is sent back to LSR1, which completes the end-to-end CR-LSP setup.
A request at ingress LSR to setup a CR-LSP might originate from a management system or an application, the details are implementation specific.
Jamoussi, et al. Standards Track [Page 34]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The ingress LSR uses information provided by the management system or the application and possibly also information from the routing database to calculate the explicit route and to create the Label Request Message.
The Label request message carries together with other necessary information an ER-TLV defining the explicitly routed path. In our example the list of hops in the ER-Hop TLV is supposed to contain an abstract node representing a group of nodes, an abstract node representing a specific node, another abstract node representing a group of nodes, and an abstract node representing a specific egress point.
In--{Group 1}--{Specific A}--{Group 2}--{Specific Out: B} The ER-TLV contains four ER-Hop TLVs:
1. An ER-Hop TLV that specifies a group of LSR valid for the first abstract node representing a group of nodes (Group 1).
2. An ER-Hop TLV that indicates the specific node (Node A).
3. An ER-Hop TLV that specifies a group of LSRs valid for the second abstract node representing a group of nodes (Group 2).
4. An ER-Hop TLV that indicates the specific egress point for the CR-LSP (Node B).
All the ER-Hop TLVs are strictly routed nodes.
The setup procedure for this CR-LSP works as follows:
1. The ingress node sends the Label Request Message to a node that is a member the group of nodes indicated in the first ER- Hop TLV, following normal routing for the specific node (A).
2. The node that receives the message identifies itself as part of the group indicated in the first ER-Hop TLV, and that it is not the specific node (A) in the second. Further it realizes that the specific node (A) is not one of its next hops.
3. It keeps the ER-Hop TLVs intact and sends a Label Request Message to another node that is part of the group indicated in the first ER-Hop TLV (Group 1), following normal routing for the specific node (A).
Jamoussi, et al. Standards Track [Page 35]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
4. The node that receives the message identifies itself as part of the group indicated in the first ER-Hop TLV, and that it is not the specific node (A) in the second ER-Hop TLV. Further it realizes that the specific node (A) is one of its next hops.
5. It removes the first ER-Hop TLVs and sends a Label Request Message to the specific node (A).
6. The specific node (A) recognizes itself in the first ER-Hop TLV. Removes the specific ER-Hop TLV.
7. It sends a Label Request Message to a node that is a member of the group (Group 2) indicated in the ER-Hop TLV.
8. The node that receives the message identifies itself as part of the group indicated in the first ER-Hop TLV, further it realizes that the specific egress node (B) is one of its next hops.
9. It sends a Label Request Message to the specific egress node (B).
10. The specific egress node (B) recognizes itself as the egress for the CR-LSP, it returns a Label Mapping Message, that will traverse the same path as the Label Request Message in the opposite direction.
Construction of an end-to-end service is the result of the rules enforced at the edge and the treatment that packets receive at the network nodes. The rules define the traffic conditioning actions that are implemented at the edge and they include policing with pass, mark, and drop capabilities. The edge rules are expected to be defined by the mutual agreements between the service providers and their customers and they will constitute an essential part of the SLA. Therefore edge rules are not included in the signaling protocol.
Packet treatment at a network node is usually referred to as the local behavior. Local behavior could be specified in many ways. One example for local behavior specification is the service frequency introduced in section 4.3.2.1, together with the resource reservation rules implemented at the nodes.
Jamoussi, et al. Standards Track [Page 36]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Edge rules and local behaviors can be viewed as the main building blocks for the end-to-end service construction. The following table illustrates the applicability of the building block approach for constructing different services including those defined for ATM.
Service PDR PBS CDR CBS EBS Service Conditioning Examples Frequency Action
DS S S =PDR =PBS 0 Frequent drop>PDR
TS S S S S 0 Unspecified drop>PDR,PBS mark>CDR,CBS
BE inf inf inf inf 0 Unspecified -
FRS S S CIR ~B_C ~B_E Unspecified drop>PDR,PBS mark>CDR,CBS,EBS
In the above table, the DS refers to a delay sensitive service where the network commits to deliver with high probability user datagrams at a rate of PDR with minimum delay and delay requirements. Datagrams in excess of PDR will be discarded.
The TS refers to a generic throughput sensitive service where the network commits to deliver with high probability user datagrams at a rate of at least CDR. The user may transmit at a rate higher than CDR but datagrams in excess of CDR would have a lower probability of being delivered.
Jamoussi, et al. Standards Track [Page 37]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
The BE is the best effort service and it implies that there are no expected service guarantees from the network.
In this scenario the customer needs to establish an LSP for supporting real-time applications such as voice and video. The Delay-sensitive (DS) service is requested in this case.
The first step is the specification of the traffic parameters in the signaling message. The two parameters of interest to the DS service are the PDR and the PBS and the user based on his requirements specifies their values. Since all the traffic parameters are included in the signaling message, appropriate values must be assigned to all of them. For DS service, the CDR and the CBS values are set equal to the PDR and the PBS respectively. An indication of whether the parameter values are subject to negotiation is flagged.
The transport characteristics of the DS service require Frequent frequency to be requested to reflect the real-time delay requirements of the service.
In addition to the transport characteristics, both the network provider and the customer need to agree on the actions enforced at the edge. The specification of those actions is expected to be a part of the service level agreement (SLA) negotiation and is not included in the signaling protocol. For DS service, the edge action is to drop packets that exceed the PDR and the PBS specifications. The signaling message will be sent in the direction of the ER path and the LSP is established following the normal LDP procedures. Each LSR applies its admission control rules. If sufficient resources are not available and the parameter values are subject to negotiation, then the LSR could negotiate down the PDR, the PBS, or both.
The new parameter values are echoed back in the Label Mapping Message. LSRs might need to re-adjust their resource reservations based on the new traffic parameter values.
In this example we assume that a throughput sensitive (TS) service is requested. For resource allocation the user assigns values for PDR, PBS, CDR, and CBS. The negotiation flag is set if the traffic parameters are subject to negotiation. Since the service is delay insensitive by definition, the Unspecified frequency is signaled to indicate that the service frequency is not an issue.
Jamoussi, et al. Standards Track [Page 38]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Similar to the previous example, the edge actions are not subject for signaling and are specified in the service level agreement between the user and the network provider.
For TS service, the edge rules might include marking to indicate high discard precedence values for all packets that exceed CDR and the CBS. The edge rules will also include dropping of packets that conform to neither PDR nor PBS.
Each LSR of the LSP is expected to run its admission control rules and negotiate traffic parameters down if sufficient resources do not exist. The new parameter values are echoed back in the Label Mapping Message. LSRs might need to re-adjust their resources based on the new traffic parameter values.
Paul Doolan On The Beach Consulting Corp 34 Mill Pond Circle Milford MA 01757 Phone 617 513 852 EMail: pdoolan@acm.org
Jamoussi, et al. Standards Track [Page 39]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Nancy Feldman IBM Research 30 Saw Mill River Road Hawthorne, NY 10532 Phone: 914-784-3254 EMail: Nkf@us.ibm.com
Andre Fredette ANF Consulting 62 Duck Pond Dr. Groton, MA 01450 EMail: afredette@charter.net
Eric Gray 600 Federal Drive Andover, MA 01810 Phone: (978) 689-1610 EMail: eric.gray@sandburst.com
Juha Heinanen Song Networks, Inc. Hallituskatu 16 33200 Tampere, Finland EMail: jh@song.fi
Bilel Jamoussi Nortel Networks 600 Technology Park Drive Billerica, MA 01821 USA Phone: +1 978 288-4506 Mail: Jamoussi@nortelnetworks.com
Timothy E. Kilty Island Consulting Phone: (978) 462 7091 EMail: tim-kilty@mediaone.net
Andrew G. Malis Vivace Networks 2730 Orchard Parkway San Jose, CA 95134 Phone: +1 408 383 7223 EMail: Andy.Malis@vivacenetworks.com
Jamoussi, et al. Standards Track [Page 40]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Muckai K Girish Atoga Systems 49026 Milmont Drive Fremont, CA 94538 EMail: muckai@atoga.com
Tom Worster Phone: 617 247 2624 EMail: fsb@thefsb.org
Liwen Wu Cisco Systems 250 Apollo Drive Chelmsford, MA. 01824 Phone: 978-244-3087 EMail: liwwu@cisco.com
Jamoussi, et al. Standards Track [Page 41]
RFC 3212 Constraint-Based LSP Setup using LDP January 2002
Full Copyright Statement
Copyright (C) The Internet Society (2002). All Rights Reserved.
This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English.
The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns.
This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Acknowledgement
Funding for the RFC Editor function is currently provided by the Internet Society.