| Internet-Draft | Flow-Spec Traffic Compress Action | November 2026 |
| Ma, et al. | Expires 20 May 2027 | [Page] |
BGP Flow Specification Version 2 (FSv2) provides an extensible framework for disseminating traffic flow specification rules and associated actions. While traditional Flow-spec actions cover rate-limiting, redirecting, and marking, emerging applications require dynamic bandwidth optimization. This document specifies a new traffic filtering action to support selective, on-demand traffic compression across network domains using BGP Extended Communities aligned with the FSv2 architecture.¶
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 20 May 2027.¶
Copyright (c) 2026 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.¶
BGP [RFC4271] Flow Specification [RFC8955] [RFC8956] and its evolution to Version 2 (FSv2) specified in [I-D.ietf-idr-fsv2-ip-basic] allow for the dissemination of traffic flow specification rules and traffic filtering actions. Traditional Flow-spec actions define widely-used filtering behaviors such as discard, rate limit, and traffic marking.¶
Data transmitted across enterprise networks, data centers, and external connections varies widely in nature (e.g., database payloads, bulk text, multimedia, and signaling). Certain traffic aggregates present significant redundancy and high compression ratios, whereas other data streams have little to no compression potential or require dedicated processing hardware.¶
While BGP Flow-spec enables network-wide policy distribution, static configurations lack the agility required for dynamic data-center interconnect (DCI) traffic engineering. Therefore, this document introduces a new BGP Flow-spec action that enables network operators to dynamically apply traffic compression based on granular traffic filters, saving core transport bandwidth and improving overall network utilization.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This document proposes a new BGP extended community called the "flow-spec traffic compress action". It operates as a Generic Transitive Extended Community type ("0x80"). The sub-type value [to be assigned by IANA] indicates that the global administrator and local administrator fields encode a flow-spec traffic compress action aligned with FSv2 action guidelines.¶
In this extended community, the 2-byte global administrator field encodes compression algorithms referred to as Compression Parameter Index (CPI) [RFC3173]. The 4-byte local administrator field carries compression parameters (CPV).¶
The new BGP extended community is encoded as follows:¶
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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | 0x80 | Sub-Type | CPI | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | CPV | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Sub-Type: TBD.¶
CPI (2 octets): The values 0-63 designate well-known compression algorithms. The values 64-255 are reserved for future use. The values 256-61439 are negotiated between the endpoints in the definition of an IPComp Association [RFC3173].¶
CPV (4 octets): Parameters associated with the selected CPI. Depending on
the CPI, the CPV can be divided into distinct segments (e.g., algorithm
tuning flags, window sizes, or application IDs). Unused bits MUST be
set to zero by the originating BGP speaker and ignored upon receipt.
For instance, when utilizing the zstd compression algorithm, the CPV can
be structured as: <trainAppId (1 octet), CRC (2 octets), appId (1 octet)>.¶
In the typical use case, the tailend node enables decompression capability and, via BGP Flow Specification, advertises the flow specification rules together with the Traffic Compress Extended Community. The CPI field of the action carries the algorithm supported by the tailend node for the associated IPComp Association.¶
Upon receiving the Flow-spec route, the headend node verifies that it locally supports compression with the algorithm indicated in the action. If the capability check succeeds, the Flow-spec rule is installed and the compression action becomes effective on the matching traffic. If the headend node does not support the indicated algorithm, the rule MUST NOT take effect, and matching traffic MUST be forwarded uncompressed; an implementation MAY record a local policy exception. No reverse notification of the check result is defined in this document.¶
The tailend node thus acts as the decompression endpoint and the headend node as the compression endpoint; the Flow-spec route propagates the compression action from the tailend node toward the headend node.¶
Flow-spec match conditions refer to the uncompressed packets. For each packet matching the Flow-spec rule that is not already compressed, the headend node compresses the payload using the algorithm indicated by the action and encapsulates the compressed packet in accordance with [RFC3173] (IP protocol number 108, IPComp). Transport mode of IPComp SHOULD be used so that the outer IP header is preserved and can still be matched by Flow-spec rules along the path. Packets already carrying IPComp encapsulation MUST be left unmodified. The tailend node identifies compressed packets by the IPComp protocol number and performs the corresponding decompression.¶
The same action may be received by more than one node, for instance when Flow-spec routes are distributed by Route Reflectors to all PEs. In such deployments, the rule above ensures that a packet compressed by an upstream node is not compressed again by an intermediate node, which simply forwards it unchanged toward the tailend node.¶
Algorithm selection is conveyed by the Flow-spec action defined in this document rather than by the CPI negotiation procedures of [RFC3173].¶
Because compression and decompression introduce processing overhead, nodes receiving this action MUST verify that they possess the hardware or software capabilities corresponding to the requested CPI. If a node lacks the requested compression capability, it MUST fall back to standard uncompressed forwarding or report a local policy exception, preventing packet dropping due to processing failures.¶
If multiple traffic compression extended communities exist in a BGP route attribute, the value of the first traffic compression extended community MUST take precedence.¶
The validation of the TLVs and extended community structures introduced in this document MUST be performed to verify formatting correctness. In case of any malformed attributes detected, error handling MUST comply with [RFC7606] and the error policies defined in [I-D.ietf-idr-fsv2-ip-basic].¶
IANA is requested to create a new entry/registry under the "Extended Community" registries:¶
SubType Description ------- --------------------------------------------- TBD Traffic Compress Extended Community¶
Traffic compression introduces potential vector risks, such as CPU exhaustion (Denial-of-Service) attacks caused by malformed or highly inefficient payload structures designed to trigger worst-case compression cycles. Operators SHOULD implement rate-limiting or resource quotas on compression-enabled forwarding engines.¶
Furthermore, standard BGP security mechanisms (such as RPKI and BGP session authentication) MUST be deployed to prevent unauthorized injection of compression action extended communities.¶
The authors would like to acknowledge the review and inputs from Haibo Wang, Gang Yan, Linda Dunbar, Zhenbin Li, Rainbow Wu, Jie Dong , Ziqing Cao, Lily Wang, Shunwan Zhuang, Gang Pei and Shuqiang Wang. The authors also thank Linda Dunbar, Joel Halpern, and Bill Fenner for their detailed review comments during the WG adoption call, which significantly improved this document.¶