|Title||RSVP Extensions for Policy Control
Network Working Group S. Herzog
Request for Comments: 2750 IPHighway
Updates: 2205 January 2000
Category: Standards Track
RSVP Extensions for Policy Control
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 (C) The Internet Society (2000). All Rights Reserved.
This memo presents a set of extensions for supporting generic policy
based admission control in RSVP. It should be perceived as an
extension to the RSVP functional specifications [RSVP]
These extensions include the standard format of POLICY_DATA objects,
and a description of RSVP's handling of policy events.
This document does not advocate particular policy control mechanisms;
however, a Router/Server Policy Protocol description for these
extensions can be found in [RAP, COPS, COPS-RSVP].
Table of Contents
2 A Simple Scenario..................................................3
3 Policy Data Objects................................................3
3.1 Base Format.....................................................4
3.3 Policy Elements.................................................7
3.4 Purging Policy State............................................7
4 Processing Rules...................................................8
4.1 Basic Signaling.................................................8
4.2 Default Handling for PIN nodes..................................8
4.3 Error Signaling.................................................9
5 IANA Considerations................................................9
6 Security Considerations............................................9
9 Author Information................................................10
Appendix A: Policy Error Codes......................................11
Appendix B: INTEGRITY computation for POLICY_DATA objects...........12
Full Copyright Statement ...........................................13
RSVP, by definition, discriminates between users, by providing some
users with better service at the expense of others. Therefore, it is
reasonable to expect that RSVP be accompanied by mechanisms for
controlling and enforcing access and usage policies. Version 1 of the
RSVP Functional Specifications [RSVP] left a placeholder for policy
support in the form of POLICY_DATA object.
The current RSVP Functional Specification describes the interface to
admission (traffic) control that is based "only" on resource
availability. In this document we describe a set of extensions to
RSVP for supporting policy based admission control as well. The scope
of this document is limited to these extensions and does not advocate
specific architectures for policy based controls.
For the purpose of this document we do not differentiate between
Policy Decision Point (PDP) and Local Decision Point (LDPs) as
described in [RAP]. The term PDP should be assumed to include LDP as
2 A Simple Scenario
It is generally assumed that policy enforcement (at least in its
initial stages) is likely to concentrate on border nodes between
Figure 1 illustrates a simple autonomous domain with two boundary
nodes (A, C) which represent PEPs controlled by PDPs. A core node (B)
represents an RSVP capable policy ignorant node (PIN) with
capabilities limited to default policy handling (Section 4.2).
+---+ +---+ +---+
| A +---------+ B +---------+ C |
+---+ +---+ +---+
PEP2 PIN PEP2
Figure 1: Autonomous Domain scenario
Here, policy objects transmitted across the domain traverse an
intermediate PIN node (B) that is allowed to process RSVP message but
considered non-trusted for handling policy information.
This document describes processing rules for both PEP as well as PIN
3 Policy Data Objects
POLICY_DATA objects are carried by RSVP messages and contain policy
information. All policy-capable nodes (at any location in the
network) can generate, modify, or remove policy objects, even when
senders or receivers do not provide, and may not even be aware of
policy data objects.
The exchange of POLICY_DATA objects between policy-capable nodes
along the data path, supports the generation of consistent end-to-end
policies. Furthermore, such policies can be successfully deployed
across multiple administrative domains when border nodes manipulate
and translate POLICY_DATA objects according to established sets of
The following extends section A.13 in [RSVP].
3.1 Base Format
o Type 1 POLICY_DATA object: Class=14, C-Type=1
| Length | POLICY_DATA | 1 |
| Data Offset | 0 (reserved) |
// Option List //
// Policy Element List //
Data Offset: 16 bits
The offset in bytes of the data portion (from the first
byte of the object header).
Reserved: 16 bits
Option List: Variable length
The list of options and their usage is defined in Section
Policy Element List: Variable length
The contents of policy elements is opaque to RSVP. See more
details in Section 3.3.
This section describes a set of options that may appear in
POLICY_DATA objects. All policy options appear as RSVP objects but
their semantic is modified when used as policy data options.
FILTER_SPEC object (list) or SCOPE object
These objects describe the set of senders associated with the
POLICY_DATA object. If none is provided, the policy information is
assumed to be associated with all the flows of the session. These two
types of objects are mutually exclusive, and cannot be mixed.
In Packed FF Resv messages, this FILTER_SPEC option provides
association between a reserved flow and its POLICY_DATA objects.
In WF or SE styles, this option preserves the original
flow/POLICY_DATA association as formed by PDPs, even across RSVP
capable PINs. Such preservation is required since PIN nodes may
change the list of reserved flows on a per-hop basis, irrespective of
legitimate Edge-to-Edge PDP policy considerations.
Last, the SCOPE object should be used to prevent "policy loops" in a
manner similar to the one described in [RSVP], Section 3.4. When PIN
nodes are part of a WF reservation path, the RSVP SCOPE object is
unable to prevent policy loops and the separate policy SCOPE object
Note: using the SCOPE option may have significant impact on scaling
and size of POLICY_DATA objects.
The RSVP_HOP object identifies the neighbor/peer policy-capable node
that constructed the policy object. When policy is enforced at border
nodes, peer policy nodes may be several RSVP hops away from each
other and the originating RSVP_HOP is the basis for the mechanism
that allows them to recognize each other and communicate safely and
If no RSVP_HOP object is present, the policy data is implicitly
assumed to have been constructed by the RSVP_HOP indicated in the
RSVP message itself (i.e., the neighboring RSVP node is policy-
A second RSVP_HOP object may follow the originating RSVP_HOP object.
This second RSVP_HOP identifies the destination policy node. This is
used to ensure the POLICY_DATA object is delivered to targeted policy
nodes. It may be used to emulate unicast delivery in multicast Path
messages. It may also help prevent using a policy object in other
parts of the network (replay attack).
On the receiving side, a policy node should ignore any POLICY_DATA
that includes a destination RSVP_HOP that doesn't match its own IP
Figure 1 (Section 2) provides an example where POLICY_DATA objects
are transmitted between boundary nodes while traversing non-secure
PIN nodes. In this scenario, the RSVP integrity mechanism becomes
ineffective since it places policy trust with intermediate PIN nodes
(which are trusted to perform RSVP signaling but not to perform
policy decisions or manipulations).
The INTEGRITY object option inside POLICY_DATA object creates direct
secure communications between non-neighboring PEPs (and their
controlling PDPs) without involving PIN nodes.
This option can be used at the discretion of PDPs, and is computed in
a manner described in Appendix B.
Policy Refresh TIME_VALUES (PRT)
The Policy Refresh TIME_VALUES (PRT) option is used to slow policy
refresh frequency for policies that have looser timing constraints
relative to RSVP. If the PRT option is present, policy refreshes can
be withheld as long as at least one refresh is sent before the policy
refresh timer expires. A minimal value for PRT is R; lower values are
assumed to be R (neither error nor warning should be triggered).
To simplify RSVP processing, time values are not based directly on
the PRT value, but on a Policy Refresh Multiplier N computed as
N=Floor(PRT/R). Refresh and cleanup rules are derived from [RSVP]
Section 3.7 assuming the refresh period for PRT POLICY DATA is R'
computed as R'=N*R. In effect, both the refresh and the state
cleanup are slowed by a factor of N).
The refresh multiplier applies to no-change periodic refreshes only
(rather than updates). For example, a policy being refreshed at time
T, T+N, T+2N,... may encounter a route change detected at T+X. In
this case, the event would force an immediate policy update and would
reset srfresh times to T+X+N, T+X+2N,...
When network nodes restart, RSVP messages between PRT policy
refreshes may be rejected since they arrive without necessary
POLICY_DATA objects. This error situation would clear with the next
periodic policy refresh or with a policy update triggered by ResvErr
or PathErr messages.
This option is especially useful to combine strong (high overhead)
and weak (low overhead) authentication certificates as policy data.
In such schemes the weak certificate can support admitting a
reservation only for a limited time, after which the strong
certificate is required.
This approach may reduce the overhead of POLICY_DATA processing.
Strong certificates could be transmitted less frequently, while weak
certificates are included in every RSVP refresh.
3.3 Policy Elements
The content of policy elements is opaque to RSVP; their internal
format is understood by policy peers e.g. an RSVP Local Decision
Point (LDP) or a Policy Decision Point (PDP) [RAP]. A registry of
policy element codepoints and their meaning is maintained by [IANA-
CONSIDERATIONS] (also see Section 5).
Policy Elements have the following format:
| Length | P-Type |
// Policy information (Opaque to RSVP) //
3.4 Purging Policy State
Policy state expires in the granularity of Policy Elements
(POLICY_DATA objects are mere containers and do not expire as such).
Policy elements expire in the exact manner and time as the RSVP state
received in the same message (see [RSVP] Section 3.7). PRT
controlled state expires N times slower (see Section 3.2).
Only one policy element of a certain P-Type can be active at any
given time. Therefore, policy elements are instantaneously replaced
when another policy element of the same P-Type is received from the
same PDP (previous or next policy RSVP_HOP). An empty policy element
of a certain P-Type is used to delete (rather than a replace) all
policy state of the same P-Type.
4 Processing Rules
These sections describe the minimal required policy processing rules
4.1 Basic Signaling
This memo mandates enforcing policy control for Path, Resv, PathErr,
and ResvErr messages only. PathTear and ResvTear are assumed not to
require policy control based on two main presumptions. First, that
Integrity verification [MD5] guarantee that the Tear is received from
the same node that sent the installed reservation, and second, that
it is functionally equivalent to that node holding-off refreshes for
4.2 Default Handling for PIN nodes
Figure 1 illustrates an example of where policy data objects traverse
PIN nodes in transit from one PEP to another.
A PIN node is required at a minimum to forward the received
POLICY_DATA objects in the appropriate outgoing messages according to
the following rules:
o POLICY_DATA objects are to be forwarded as is, without any
o Multicast merging (splitting) nodes:
In the upstream direction:
When multiple POLICY_DATA objects arrive from downstream, the
RSVP node should concatenate all of them (as a list of the
original POLICY_DATA objects) and forward them with the
outgoing (upstream) message.
On the downstream direction:
When a single incoming POLICY_DATA object arrives from
upstream, it should be forwarded (copied) to all downstream
branches of the multicast tree.
The same rules apply to unrecognized policies (sub-objects) within
the POLICY_DATA object. However, since this can only occur in a
policy-capable node, it is the responsibility of the PDP and not
4.3 Error Signaling
Policy errors are reported by either ResvErr or PathErr messages with
a policy failure error code in the ERROR_SPEC object. Policy error
message must include a POLICY_DATA object; the object contains
details of the error type and reason in a P-Type specific format (See
If a multicast reservation fails due to policy reasons, RSVP should
not attempt to discover which reservation caused the failure (as it
would do for Blockade State). Instead, it should attempt to deliver
the policy ResvErr to ALL downstream hops, and have the PDP (or LDP)
decide where messages should be sent. This mechanism allows the PDP
to limit the error distribution by deciding which "culprit" next-hops
should be informed. It also allows the PDP to prevent further
distribution of ResvErr or PathErr messages by performing local
repair (e.g. substituting the failed POLICY_DATA object with a
Error codes are described in Appendix Appendix A.
5 IANA Considerations
RSVP Policy Elements (P-Types)
Following the policies outlined in [IANA-CONSIDERATIONS],numbers
0-49151 are allocated as standard policy elements by IETF Consensus
action, numbers in the range 49152-53247 are allocated as vendor
specific (one per vendor) by First Come First Serve, and numbers
53248-65535 are reserved for private use and are not assigned by
6 Security Considerations
This memo describes the use of POLICY_DATA objects to carry policy-
related information between RSVP nodes. Two security mechanisms can
be optionally used to ensure the integrity of the carried
information. The first mechanism relies on RSVP integrity [MD5] to
provide a chain of trust when all RSVP nodes are policy capable. The
second mechanism relies on the INTEGRITY object within the
POLICY_DATA object to guarantee integrity between non-neighboring
RSVP PEPs (see Sections 2 and 3.2).
[RAP] Yavatkar, R., Pendarakis, D. and R. Guerin, "A
Framework for Policy Based Admission Control",
RFC 2753, January 2000.
[COPS] Boyle, J., Cohen, R., Durham, D., Herzog, S.,
Raja, R. and A. Sastry, "The COPS (Common Open
Policy Service) Protocol", RFC 2748, January
[COPS-RSVP] Boyle, J., Cohen, R., Durham, D., Herzog, S.,
Raja, R. and A. Sastry, "COPS Usage for RSVP",
RFC 2749, January 2000.
[RSVP] Braden, R., Ed., Zhang, L., Berson, S., Herzog,
S. and S. Jamin, "Resource ReSerVation Protocol
(RSVP) - Functional Specification", RFC 2205,
[MD5] Baker, F., Lindell B. and M. Talwar, "RSVP
Cryptographic Authentication", RFC 2747,
[IANA-CONSIDERATIONS] Alvestrand, H. and T. Narten, "Guidelines for
Writing an IANA Considerations Section in
RFCs", BCP 26, RFC 2434, October 1998.
This document incorporates inputs from Lou Berger, Bob Braden,
Deborah Estrin, Roch Guerin, Timothy O'Malley, Dimitrios Pendarakis,
Raju Rajan, Scott Shenker, Andrew Smith, Raj Yavatkar, and many
9 Author Information
55 New York Avenue
Framingham, MA 01701
Phone: (508) 620-1141
Appendix A: Policy Error Codes
This Appendix extends the list of error codes described in Appendix B
Note that Policy Element specific errors are reported as described in
Section 4.3 and cannot be reported through RSVP (using this
mechanism). However, this mechanism provides a simple, less secure
mechanism for reporting generic policy errors. Most likely the two
would be used in concert such that a generic error code is provided
by RSVP, while Policy Element specific errors are encapsulated in a
return POLICY_DATA object (as in Section 4.3).
ERROR_SPEC class = 6
Error Code = 02: Policy Control failure
Error Value: 16 bit
0 = ERR_INFO : Information reporting
1 = ERR_WARN : Warning
2 = ERR_UNKNOWN : Reason unknown
3 = ERR_REJECT : Generic Policy Rejection
4 = ERR_EXCEED : Quota or Accounting violation
5 = ERR_PREEMPT : Flow was preempted
6 = ERR_EXPIRED : Previously installed policy expired (not
7 = ERR_REPLACED: Previous policy data was replaced & caused
8 = ERR_MERGE : Policies could not be merged (multicast)
9 = ERR_PDP : PDP down or non functioning
10= ERR_SERVER : Third Party Server (e.g., Kerberos) unavailable
11= ERR_PD_SYNTX: POLICY_DATA object has bad syntax
12= ERR_PD_INTGR: POLICY_DATA object failed Integrity Check
13= ERR_PE_BAD : POLICY_ELEMENT object has bad syntax
14= ERR_PD_MISS : Mandatory PE Missing (Empty PE is in the PD
15= ERR_NO_RSC : PEP Out of resources to handle policies.
16= ERR_RSVP : PDP encountered bad RSVP objects or syntax
17= ERR_SERVICE : Service type was rejected
18= ERR_STYLE : Reservation Style was rejected
19= ERR_FL_SPEC : FlowSpec was rejected (too large)
Values between 2^15 and 2^16-1 can be used for site and/or vendor
Appendix B: INTEGRITY computation for POLICY_DATA objects
Computation of the INTEGRITY option is based on the rules set forth
in [MD5], with the following modifications:
Rather than computing digest for an RSVP message, a digest is
computed for a POLICY_DATA object in the following manner:
(1) The INTEGRITY object is inserted in the appropriate place in
the POLICY_DATA object, and its location in the message is
remembered for later use.
(2) The PDP, at its discretion, and based on destination PEP/PDP
or other criteria, selects an Authentication Key and the hash
algorithm to be used.
(3) A copy of RSVP SESSION object is temporarily appended to the
end of the PD object (for the computation purposes only,
without changing the length of the POLICY_DATA object). The
flags field of the SESSION object is set to 0. This
concatenation is considered as the message for which a digest
is to be computed.
(4) The rest of the steps in Section 4.1 ((4)..(9)) remain
unchanged when computed over the concatenated message.
Note: When the computation is complete, the SESSION object is ignored
and is not part of the POLICY_DATA object.
The processing of a received POLICY_DATA object as well as a
challenge-response INTEGRITY object inside a POLICY_DATA object is
performed in the manner described in [MD5]. This processing is
subject to the modified computation algorithm as described in the
beginning of this appendix (for Section 4.1 of [MD5]).
Full Copyright Statement
Copyright (C) The Internet Society (2000). 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
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.
Funding for the RFC Editor function is currently provided by the