Skip to main content

Gap Analysis, Problem Statement, and Requirements for Inter-Domain SAV
draft-ietf-savnet-inter-domain-problem-statement-12

Document Type Active Internet-Draft (savnet WG)
Authors Dan Li , Lancheng Qin , Libin Liu , Mingqing(Michael) Huang , Kotikalapudi Sriram
Last updated 2025-10-20
Replaces draft-wu-savnet-inter-domain-problem-statement
RFC stream Internet Engineering Task Force (IETF)
Intended RFC status (None)
Formats
Additional resources Mailing list discussion
Stream WG state WG Document
Document shepherd (None)
IESG IESG state I-D Exists
Consensus boilerplate Unknown
Telechat date (None)
Responsible AD (None)
Send notices to (None)
draft-ietf-savnet-inter-domain-problem-statement-12
Internet Engineering Task Force                                    D. Li
Internet-Draft                                       Tsinghua University
Intended status: Informational                                    L. Qin
Expires: 23 April 2026                                            L. Liu
                                                 Zhongguancun Laboratory
                                                                M. Huang
                                                                  Huawei
                                                               K. Sriram
                                                                USA NIST
                                                         20 October 2025

 Gap Analysis, Problem Statement, and Requirements for Inter-Domain SAV
          draft-ietf-savnet-inter-domain-problem-statement-12

Abstract

   This document provides a gap analysis of existing inter-domain source
   address validation mechanisms, describes the problem space, and
   defines the requirements for technical improvements.

Status of This Memo

   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 23 April 2026.

Copyright Notice

   Copyright (c) 2025 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

Li, et al.                Expires 23 April 2026                 [Page 1]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   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.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
     1.1.  Requirements Language . . . . . . . . . . . . . . . . . .   4
   2.  Terminology . . . . . . . . . . . . . . . . . . . . . . . . .   4
   3.  Existing Inter-domain SAV Mechanisms  . . . . . . . . . . . .   5
   4.  Gap Analysis  . . . . . . . . . . . . . . . . . . . . . . . .   7
     4.1.  SAV at Customer Interfaces  . . . . . . . . . . . . . . .   7
       4.1.1.  Limited Propagation of Prefixes . . . . . . . . . . .   9
       4.1.2.  Hidden Prefixes . . . . . . . . . . . . . . . . . . .  10
       4.1.3.  Source Address Spoofing within a Customer Cone  . . .  12
     4.2.  SAV at Provider/Peer Interfaces . . . . . . . . . . . . .  13
   5.  Problem Statement . . . . . . . . . . . . . . . . . . . . . .  16
   6.  Requirements for New Inter-domain SAV Mechanisms  . . . . . .  18
     6.1.  Accurate Validation . . . . . . . . . . . . . . . . . . .  18
       6.1.1.  Improving Validation Accuracy over Existing
               Mechanisms  . . . . . . . . . . . . . . . . . . . . .  18
       6.1.2.  Working in Incremental/Partial Deployment . . . . . .  19
       6.1.3.  Providing Necessary Security Guarantee  . . . . . . .  20
     6.2.  Automatic Update  . . . . . . . . . . . . . . . . . . . .  20
       6.2.1.  Reducing Operational Overhead . . . . . . . . . . . .  20
       6.2.2.  Guaranteeing Convergence  . . . . . . . . . . . . . .  20
   7.  Inter-domain SAV Scope  . . . . . . . . . . . . . . . . . . .  20
   8.  Security Considerations . . . . . . . . . . . . . . . . . . .  21
   9.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .  21
   10. Contributors  . . . . . . . . . . . . . . . . . . . . . . . .  21
   11. References  . . . . . . . . . . . . . . . . . . . . . . . . .  21
     11.1.  Normative References . . . . . . . . . . . . . . . . . .  21
     11.2.  Informative References . . . . . . . . . . . . . . . . .  22
   Acknowledgements  . . . . . . . . . . . . . . . . . . . . . . . .  23
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  23

1.  Introduction

   Source address validation (SAV) is crucial for protecting networks
   from source address (SA) spoofing attacks [RFC2827] [RFC3704]
   [RFC8704].  The MANRS initiative advocates deploying SAV as close to
   the source as possible [manrs], and access networks are the first
   line of defense against source address spoofing.  However, access
   networks face various challenges in deploying SAV mechanisms due to
   different network environments, router vendors, and operational
   preferences.  Hence, SAV may not be deployed ubiquitously in access
   networks.  In addition, SA spoofing may also originate in ISP
   networks at higher levels of hierarchy in the Internet.  So,

Li, et al.                Expires 23 April 2026                 [Page 2]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   deployment of SAV mechanisms in the edge routers of enterprises as
   well as the ISP networks (at different hierarchichal levels or tiers)
   is needed to prevent source address spoofing along the data
   forwarding paths.  [RFC5210] highlights the importance of SAV at
   various network locations: access, intra-domain (intra-AS), and
   inter-domain (inter-AS).  This document focuses on providing gap
   analysis and describing the problem space of existing inter-domain
   SAV solutions, and defining the requirements for a new solution of
   these problems.  Access Control List (ACL) and unicast Reverse Path
   Forwarding (uRPF) techniques are currently utilized for inter-domain
   SAV [RFC3704] [RFC8704].  Here only existing IETF RFCs are considered
   as the state of the art (BCP 38 [RFC2827] and BCP 84 [RFC3704]
   [RFC8704]); IETF works-in-progress are not included in that.

   The terms customer, provider (transit provider), and lateral peer
   (non-transit peer; peer (for simplicity)) used in this document are
   consistent with those defined in [RFC7908] [RFC9234].  Further,
   [RFC9234] mentions Route Server (RS) and RS-client.  An RS-to-RS-
   client interface is akin to a customer interface, and an RS-client-
   to-RS interface is akin to a provider interface for the purposes of
   SAV.

   There are several existing mechanisms for inter-domain SAV.  This
   document analyzes them and attempts to answer: i) what are the
   technical gaps (Section 4), ii) what are the fundamental problems
   (Section 5), and iii) what are the practical requirements for the
   solution of these problems (Section 6).

   The following summarizes the fundamental problems with existing SAV
   mechanisms, as analyzed in Section 4 and Section 5:

   *  Improper block: Existing uRPF-based mechanisms suffer from
      improper block (false positives) in two inter-domain scenarios:
      limited propagation of prefixes and hidden prefixes.

   *  Improper permit: With some existing uRPF-based SAV mechanisms,
      improper permit (false negatives) can happen on any type interface
      (customer, lateral peer, or provider).  Specifically, if the
      method relaxes the directionality constraint [RFC3704] [RFC8704]}
      to try to achieve zero improper blocking, the possibility of
      improper permit increases.  (Note: It is recognized that unless
      there is full adoption of SAV in the customer cone (CC) of the
      interface in consideration, improper permit is not fully prevented
      in scenarios where source address spoofing occurs from within the
      CC, i.e., a prefix at one Autonomous System (AS) in the CC is
      spoofed from another AS in the CC.)

Li, et al.                Expires 23 April 2026                 [Page 3]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   *  High operational overhead: ACL-based ingress SAV filtering
      introduces significant operational overhead, as it needs to update
      ACL rules manually to adapt to prefix or routing changes in a
      timely manner.

   To address these problems, in Section 6, this document outlines the
   following technical requirements for a new solution:

   *  Improved SAV accuracy over existing mechanisms: A new solution
      MUST avoid improper block and minimize improper permit.

   *  Reduced operational overhead: A new solution MUST have less
      operational overhead than ACL-based ingress SAV filtering.

   In addition, this document defines three more requirements to ensure
   practicality:

   *  Benefit in incremental/partial deployment: A new solution MUST NOT
      assume pervasive adoption including the adoption of both SAV
      method and SAV-related information (e.g., RPKI object
      registrations) and SHOULD provide benefit to early adopters in
      terms of effective SAV performance.

   *  Providing necessary security guarantee: For solutions involving
      communication of new SAV-specific information between ASes, there
      SHOULD be a security mechanism assuring trustworthiness of the
      information.

   *  Efficient convergence: A new solution SHOULD achieve efficient SAV
      rule (SAV table) convergence in response to prefix or route
      changes.

1.1.  Requirements Language

   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.

2.  Terminology

   SAV Rule:
      The rule that indicates the validity of a specific source IP
      address or source IP prefix.

   Improper Block:

Li, et al.                Expires 23 April 2026                 [Page 4]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

      The validation results that the packets with legitimate source
      addresses are blocked improperly due to inaccurate SAV rules.

   Improper Permit:
      The validation results that the packets with spoofed source
      addresses are permitted improperly due to inaccurate SAV rules.

3.  Existing Inter-domain SAV Mechanisms

   Inter-domain SAV is typically performed at the AS level (on a per
   neighbor-AS-interface basis) and can be deployed at AS border routers
   (ASBRs) to prevent source address spoofing.  There are various
   mechanisms available to implement inter-domain SAV for anti-spoofing
   ingress filtering [nist] [manrs] [isoc], which are reviewed in this
   section.

   *  ACL-based ingress filtering [RFC3704]: ACL-based ingress SAV
      filtering is a technique that relies on ACL rules to filter
      packets based on their source addresses.  It can be applied at
      provider interfaces, lateral peer interfaces, or customer
      interfaces of an AS, and is recommended for deployment at provider
      interfaces [manrs].  At the provider interfaces, ACL-based ingress
      SAV filtering can block source prefixes that are clearly invalid
      in the inter-domain routing context, such as IANA special purpose
      or unallocated IPv4/IPv6 prefixes and the AS's internal-only
      prefixes.  However, ACL-based ingress SAV filtering introduces
      significant operational overhead, as ACL rules need to be updated
      in a timely manner to reflect prefix or routing changes in the
      inter-domain routing system.  It is also impractical to store a
      very large and dynamically varying unallocated IPv6 prefixes.  At
      the customer interfaces, ACL-based ingress filtering is less
      desirable.  Other techniques (as described below) are more
      effective for ingress SAV filtering on customer interfaces.  ACL-
      based ingress SAV filtering has applicability for broadband cable
      or digital subscriber access loop (DSL) access networks where the
      service provider has clear knowledge of IP address prefixes it has
      allocated to manage those services.

   *  uRPF-based mechanisms: A class of SAV mechanisms are based on
      Unicast Reverse Path Forwarding (uRPF) [RFC3704].  The core idea
      of uRPF for SAV is to exploit the symmetry of inter-domain
      routing: in many cases, the best next hop for a destination is
      also the best previous hop for the source.  In other words, if a
      packet arrives from a certain interface, the source address of
      that packet should be reachable via the same interface, according
      to the FIB.  However, symmetry in routing does not always holds in
      practice, and to address cases where it does not hold, many
      enhancements and modes of uRPF are proposed.  Different modes of

Li, et al.                Expires 23 April 2026                 [Page 5]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

      uRPF have different levels of strictness and flexibility, and
      network operators can choose from them to suit particular network
      scenarios.  We describe these modes as follows:

      -  Strict uRPF [RFC3704]: Strict uRPF is the most stringent mode,
         and it only permits packets that have a source address that is
         covered by a prefix in the FIB, and that the next hop for that
         prefix is the same as the incoming interface.  This mode is
         recommended for deployment at customer interfaces that directly
         connect to an AS with suballocated address space, as it can
         prevent spoofing attacks from that AS or its downstream ASes
         [nist].

      -  Loose uRPF [RFC3704]: Loose uRPF verifies that the source
         address of the packet is routable in the Internet by matching
         it with one or more prefixes in the FIB, regardless of which
         interface the packet arrives at.  If the source address is not
         routable, Loose uRPF discards the packet.  Loose uRPF is
         typically deployed at the provider interfaces of an AS to block
         packets with source addresses in prefixes that are not routed
         in the global Internet (e.g., IANA allocated private-use
         addresses, multicast addresses, etc.) [nist].

      -  Feasible Path uRPF (FP-uRPF) [RFC3704]: maintains a reverse
         path forwarding (RPF) list, which contains the prefixes and all
         their permissible routes including the optimal and alternative
         ones.  It permits an incoming packet only if the packet's
         source address is encompassed in the prefixes of the RPF list
         and its incoming interface is included in the permissible
         routes of the corresponding prefix.  FP-uRPF is recommended to
         be deployed at customer interfaces or lateral peer interfaces,
         especially those that are connected to multi-homed customer
         ASes [nist].

      -  Virtual routing and forwarding (VRF) uRPF [RFC4364] [urpf]
         [manrs]: VRF uRPF uses a separate VRF table for each external
         BGP peer and is only a way of implementation for a SAV table.

      -  Enhanced Feasible Path uRPF (EFP-uRPF) [RFC8704]: EFP-uRPF is
         based on the principle that if BGP updates for multiple
         prefixes with the same origin AS were received on different
         interfaces (at border routers), then incoming data packets with
         source addresses in any of those prefixes should be accepted on
         any of those interfaces.  The EFP-uRPF specification provides
         two alternate algorithms: Algorithm A which is stricter with a
         greater sense of directionality and Algorithm B which is more
         permissive with a lesser sense of directionality.  EFP-uRPF can
         more effectively accommodate asymmetric routing scenarios than

Li, et al.                Expires 23 April 2026                 [Page 6]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

         FP-uRPF.  EFP-uRPF is a part of BCP84.  EFP-uRPF can be used at
         customer as well as lateral peer interfaces of an AS.  It is
         not deployed yet in the Internet.

   *  Carrier Grade NAT (CGN): CGN is a network technology used by
      service providers to translate between private and public IPv4
      addresses within their network.  CGN enables service providers to
      assign private IPv4 addresses to their customer ASes instead of
      public, globally unique IPv4 addresses.  The private side of the
      CGN faces the customer ASes, and when an incoming packet is
      received from a customer AS, CGN checks its source address.  If
      the source address is included in the address list of the CGN's
      private side, CGN performs address translation.  Otherwise, it
      forwards the packet without translation.  However, since CGN
      cannot determine whether the source address of an incoming packet
      is spoofed or not, additional SAV mechanisms need to be
      implemented to prevent source address spoofing [manrs].

4.  Gap Analysis

   Inter-domain SAV is essential in preventing source address spoofing
   traffic across all AS interfaces, including those of customers,
   providers, and lateral peers.  An ideal inter-domain SAV mechanism
   MUST block all spoofing traffic while permitting legitimate traffic
   in all scenarios.  However, in some cases, existing SAV mechanisms
   may unintentionally block legitimate traffic or permit spoofing
   traffic.  This section aims to conduct a gap analysis of existing SAV
   mechanisms used in the corresponding interfaces of these scenarios to
   identify their technical limitations.

4.1.  SAV at Customer Interfaces

   SAV is used at customer interfaces to validate traffic from the
   customer cone, including both legitimate traffic and spoofing
   traffic.  To prevent the source address spoofing, operators can
   enable ACL-based ingress filtering and/or uRPF-based mechanisms at
   customer interfaces, namely Strict uRPF, FP-uRPF, or EFP-uRPF.
   However, uRPF-based mechanisms may cause improper block problems in
   two inter-domain scenarios: limited propagation of prefixes and
   hidden prefixes, or may cause improper permit problems in the
   scenarios of source address spoofing within a customer cone, while
   ACL-based SAV ingress filtering needs to update SAV rules in a timely
   manner and lead to high operational overhead.

Li, et al.                Expires 23 April 2026                 [Page 7]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   +--------------------+------------+-----------+-------+--------+
   |Traffic & Scenarios |     ACL    |Strict uRPF|FP-uRPF|EFP-uRPF|
   +----------+---------+------------+-----------+-------+--------+
   |Legitimate|   LPP   |            |                            |
   |Traffic   +---------+            |       Improper Block       |
   |          |   HP    |    High    |                            |
   +----------+---------+Operational-+-------------------+--------+
   |          |         |  Overhead  |                   |Improper|
   |Spoofed   |  no SCC |            |                   |Permit  |
   |Traffic   |         |            |   Functions as    |only for|
   |          |         |            |      Expected     |EFP-uRPF|
   |          |         |            |                   |Alg. B  |
   |+---------+---------+            +-------------------+--------|
   |Spoofed   |   SCC   |            |                            |
   |Traffic   |         |            |       Improper Permit      |
   |          |         |            |                            |
   +----------+---------+------------+----------------------------+

   "LPP" is a scenario with limited propagation of a prefix
   (e.g., due to NO_EXPORT Community {{noexp}}).
   "HP" is a scenario with hidden prefixes (e.g., Direct Server
   Return (DSR) {{dsrp}} anycast prefix.)
   "SCC" refers to Spoofing within a CC. It is a scenario where
   in the customer cone (CC) of the interface in consideration,
   source address spoofing occurs from within the CC,
   i.e., a prefix at one AS in the CC is spoofed from
   another AS in the CC.
   "Functions as Expected" can be used to connote that the
   inter-domain SAV mechanism does not cause improper permit for
   spoofed traffic in the scenario under consideration.
   It can also be used to connote that the SAV mechanism
   does not cause improper block for legitimate traffic in the
   scenario under consideration. In both cases, it also means that
   the SAV mechanism has low operational overhead.

      Figure 1: The gaps of ACL-based ingress filtering, Strict uRPF,
            FP-uRPF, and EFP-uRPF in the scenarios of interest.

   Figure 1 provides an overview of the gaps associated with ACL-based
   ingress filtering, Strict uRPF, FP-uRPF, and EFP-uRPF for SAV at
   customer interfaces in the corresponding scenarios.  ACL-based
   ingress filtering has high operational overhead as performing SAV at
   customer interfaces.  Strict uRPF, FP-uRPF, and EFP-uRPF, on the
   other hand, may incorrectly block legitimate traffic in the scenarios
   of limited propagation of prefixes or hidden prefixes.  Furthermore,
   in the scenarios of source address spoofing within a customer cone,
   EFP-uRPF with algorithm B may inadvertently permit the spoofing
   traffic.

Li, et al.                Expires 23 April 2026                 [Page 8]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   In the following, we analyze the gaps of Strict uRPF, FP-uRPF, and
   EFP-uRPF for SAV at customer interfaces in scenarios of limited
   propagation of prefixes, hidden prefixes, and source address spoofing
   within a customer cone, respectively.

4.1.1.  Limited Propagation of Prefixes

   In inter-domain networks, some prefixes may not be propagated to all
   domains due to various factors, such as NO_EXPORT or NO_ADVERTISE
   communities or other route filtering policies.  This may cause
   asymmetric routing in the inter-domain context, which may lead to
   improper block when performing SAV with existing mechanisms.  These
   mechanisms include EFP-uRPF, which we focus on in the following
   analysis, as well as Strict uRPF and FP-uRPF.  All these mechanisms
   suffer from the same problem of improper block in this scenario.

                             +----------------+
                             |    AS 3(P3)    |
                             +-+/\------+/\+--+
                                /         \
                               /           \
                              /             \
                             / (C2P)         \
                    +------------------+      \
                    |     AS 4(P4)     |       \
                    ++/\+--+/\+----+/\++        \
                      /     |        \           \
            P2[AS 2] /      |         \           \
                    /       |          \           \
                   / (C2P)  |           \ P5[AS 5]  \ P5[AS 5]
   +----------------+       |            \           \
   |    AS 2(P2)    |       | P1[AS 1]    \           \
   +----------+/\+--+       | P6[AS 1]     \           \
                \           |               \           \
        P1[AS 1] \          |                \           \
        NO_EXPORT \         |                 \           \
                   \ (C2P)  | (C2P/P2P)  (C2P) \     (C2P) \
                 +----------------+          +----------------+
                 |  AS 1(P1, P6)  |          |    AS 5(P5)    |
                 +----------------+          +----------------+

       Figure 2: Limited propagation of prefixes caused by NO_EXPORT.

   Figure 2 presents a scenario where the limited propagation of
   prefixes occurs due to the NO_EXPORT community attribute.  In this
   scenario, AS 1 is a customer of AS 2, AS 2 is a customer of AS 4, AS
   4 is a customer of AS 3, and AS 5 is a customer of both AS 3 and AS
   4.  The relationship between AS 1 and AS 4 can be either customer-to-

Li, et al.                Expires 23 April 2026                 [Page 9]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   provider (C2P) or peer-to-peer (P2P).  AS 1 advertises prefixes P1 to
   AS 2 and adds the NO_EXPORT community attribute to the BGP
   advertisement sent to AS 2, preventing AS 2 from further propagating
   the route for prefix P1 to AS 4.  Consequently, AS 4 only learns the
   route for prefix P1 from AS 1 in this scenario.  Suppose AS 1 and AS
   4 have deployed inter-domain SAV while other ASes have not, and AS 4
   has deployed EFP-uRPF at its customer interfaces.

   Assuming that AS 1 is the customer of AS 4, if AS 4 deploys EFP-uRPF
   with algorithm A at customer interfaces, it will require packets with
   source addresses in P1 or P6 to only arrive from AS 1.  When AS 1
   sends legitimate packets with source addresses in P1 or P6 to AS 4
   through AS 2, AS 4 improperly blocks these packets.  The same problem
   applies to Strict uRPF and FP-uRPF.  EFP-uRPF with algorithm B can
   avoid improper block in this case in principle.

4.1.2.  Hidden Prefixes

   Some servers' source addresses are not advertised through BGP to
   other ASes.  These addresses are unknown to the inter-domain routing
   system and are called hidden prefixes.  Legitimate traffic using
   these hidden prefixes as source addresses would be dropped by
   existing inter-domain SAV mechanisms, such as Strict uRPF, FP-uRPF,
   or EFP-uRPF, because they do not match any known prefix.

   For example, Content Delivery Networks (CDN) use anycast [RFC4786]
   [RFC7094] to improve the quality of service by bringing content
   closer to users.  An anycast IP address is assigned to devices in
   different locations, and incoming requests are routed to the closest
   location.  Usually, only locations with multiple connectivity
   announce the anycast IP address through BGP.  The CDN server receives
   requests from users and creates tunnels to the edge locations, where
   content is sent directly to users using direct server return (DSR).
   DSR requires servers in the edge locations to use the anycast IP
   address as the source address in response packets.  However, these
   edge locations do not announce the anycast prefixes through BGP, so
   an intermediate AS with existing inter-domain SAV mechanisms may
   improperly block these response packets.

Li, et al.                Expires 23 April 2026                [Page 10]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

                                   +----------------+
                   Anycast Server+-+    AS 3(P3)    |
                                   +-+/\----+/\+----+
                                      /       \
                            P3[AS 3] /         \
                                    /           \
                                   / (C2P)       \
                          +----------------+      \
                          |    AS 4(P4)    |       \
                          ++/\+--+/\+--+/\++        \
             P6[AS 2, AS 1] /     |      \           \
            P1[AS 2, AS 1] /      |       \           \
                 P2[AS 2] /       |        \           \
                         / (C2P)  |         \ P5[AS 5]  \ P5[AS 5]
         +----------------+       |          \           \
   User+-+    AS 2(P2)    |       | P1[AS 1]  \           \
         +----------+/\+--+       | P6[AS 1]   \           \
                      \           |             \           \
              P6[AS 1] \          |              \           \
               P1[AS 1] \         |               \           \
                         \ (C2P)  | (C2P)    (C2P) \     (C2P) \
                       +----------------+        +----------------+
          Edge Server+-+  AS 1(P1, P6)  |        |    AS 5(P5)    |
                       +----------------+        +----------------+
   P3 is the anycast prefix and is only advertised by AS 3 through BGP.

              Figure 3: A Direct Server Return (DSR) scenario.

   Figure 3 illustrates a DSR scenario where the anycast IP prefix P3 is
   only advertised by AS 3 through BGP.  In this example, AS 3 is the
   provider of AS 4 and AS 5, AS 4 is the provider of AS 1, AS 2, and AS
   5, and AS 2 is the provider of AS 1.  AS 4 has deployed inter-domain
   SAV.  When users in AS 2 send requests to the anycast destination IP,
   the forwarding path is AS 2->AS 4->AS 3.  The anycast servers in AS 3
   receive the requests and tunnel them to the edge servers in AS 1.
   Finally, the edge servers send the content to the users with source
   addresses in prefix P3.  The reverse forwarding path is AS 1->AS
   4->AS 2.  Since AS 4 does not receive routing information for prefix
   P3 from AS 1, EFP-uRPF with algorithm A/B, and all other existing
   uRPF-based mechanisms at the customer interface of AS 4 facing AS 1
   will improperly block the legitimate response packets from AS 1.

   Further, in some network scenarios, such as multicast and satellite
   networks, specific prefixes may be exclusively used as source
   addresses without being advertised via BGP by any AS.  While
   different from DSR scenarios, these cases similarly result in
   existing inter-domain SAV mechanisms improperly blocking legitimate
   traffic originating from such prefixes.

Li, et al.                Expires 23 April 2026                [Page 11]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

4.1.3.  Source Address Spoofing within a Customer Cone

   EFP-uRPF with algorithm B may also permit spoofing traffic improperly
   in scenarios where source address spoofing within a customer cone
   occur.  We provide illustrations of these scenarios using an example
   in the following.  The source address spoofing within a customer cone
   represents a class of scenario where spoofing traffic comes from a
   customer AS within a customer cone and the spoofed source addresses
   belong to this customer cone.  Figure 4 portrays a scenario of source
   address spoofing within a customer cone and is used to analyze the
   gaps of uRPF-based mechanisms.

                                        +----------------+
                                        |    AS 3(P3)    |
                                        +-+/\----+/\+----+
                                           /       \
                                          /         \
                                         /           \
                                        / (C2P)       \
                               +----------------+      \
                               |    AS 4(P4)    |       \
                               ++/\+--+/\+--+/\++        \
                  P6[AS 2, AS 1] /     |      \           \
                 P1[AS 2, AS 1] /      |       \           \
                      P2[AS 2] /       |        \           \
                              / (C2P)  |         \ P5[AS 5]  \ P5[AS 5]
              +----------------+       |          \           \
 Spoofer(P5')-+    AS 2(P2)    |       | P1[AS 1]  \           \
              +----------+/\+--+       | P6[AS 1]   \           \
                           \           |             \           \
                   P6[AS 1] \          |              \           \
                    P1[AS 1] \         |               \           \
                              \ (C2P)  | (C2P)    (C2P) \     (C2P) \
                            +----------------+        +----------------+
                            |  AS 1(P1, P6)  |        |    AS 5(P5)    |
                            +----------------+        +----------------+
 P5' is the spoofed source prefix P5 by the spoofer which is inside of
 AS 2 or connected to AS 2 through other ASes.

   Figure 4: A scenario of source address spoofing within a customer
                                 cone.

   In Figure 4, the source address spoofing takes place within AS 4's
   customer cone, where the spoofer, which is inside of AS 2 or
   connected to AS 2 through other ASes, sends spoofing traffic with
   spoofed source addresses in P5 to AS 3 along the path AS 2->AS 4-> AS
   3.  The arrows in Figure 4 illustrate the commercial relationships
   between ASes.  AS 3 serves as the provider for AS 4 and AS 5, while

Li, et al.                Expires 23 April 2026                [Page 12]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   AS 4 acts as the provider for AS 1, AS 2, and AS 5.  Additionally, AS
   2 is the provider for AS 1.  Suppose AS 1 and AS 4 have deployed
   inter-domain SAV, while the other ASes have not.

   If AS 4 deploys EFP-uRPF with algorithm B at its customer interfaces,
   it will allow packets with source addresses in P5 to originate from
   AS 1, AS 2, and AS 5.  Consequently, when the spoofer which is inside
   of AS 2 or connected to AS 2 through other ASes sends spoofing
   packets with spoofed source addresses in P5 to AS 3, AS 4 will
   improperly permit these packets, thus enabling the spoofing traffic
   to propagate.

   In scenarios like these, Strict uRPF, FP-uRPF, VRF uRPF, and EFP-uRPF
   with algorithm A do not suffer from improper permit problems.  This
   is because these mechanisms enforce strict filtering rules that
   ensure packets with source addresses in P5 are only permitted to
   arrive at AS 4's customer interfaces facing AS 5.

4.2.  SAV at Provider/Peer Interfaces

   SAV is used at provider/peer interfaces to validate traffic entering
   the customer cone, including both legitimate and spoofing traffic.
   To prevent packets with spoofed source addresses from the provider/
   peer AS, ACL-based ingress filtering and/or Loose uRPF can be
   deployed [nist].

   +------------------------+------------+---------------+
   |   Traffic & Scenarios  |     ACL    |   Loose uRPF  |
   +----------+-------------+------------+---------------+
   |Legitimate|      Any    |            |  Functions    |
   |Traffic   |  Scenarios  |    High    |  as Expected  |
   +----------+-------------+Operational +---------------+
   |Spoofed   |   Spoofed   |  Overhead  |               |
   |Traffic   |     from    |            |Improper Permit|
   |          |provider/peer|            |               |
   |          |      AS     |            |               |
   +----------+-------------+------------+---------------+
   "Spoofed from provider/peer AS" is a scenario where the
   spoofed traffic comes from a provider/peer AS and the spoofed
   prefix belongs to the customer cone of the AS that
   is performing SAV.
   "Functions as Expected" can be used to connote that the
   inter-domain SAV mechanism does not cause improper permit for
   spoofed traffic in the scenario under consideration.
   It can also be used to connote that the SAV mechanism
   does not cause improper block for legitimate traffic in the
   scenario under consideration. In both cases, it also means that
   the SAV mechanism has low operational overhead.

Li, et al.                Expires 23 April 2026                [Page 13]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

      Figure 5: The gaps of ACL-based ingress filtering and Loose uRPF
         at Provider/ Peer Interfaces in the scenarios of interest.

   Figure 5 summarizes the gaps of ACL-based ingress filtering and Loose
   uRPF for SAV at provider/peer interfaces in the corresponding
   scenarios.  ACL-based ingress filtering effectively blocks spoofing
   traffic from provider/peer AS, while appropriately allowing
   legitimate traffic.  However, these methods may come with high
   operational overhead.  On the other hand, Loose uRPF correctly
   permits legitimate traffic, but it can also mistakenly allow spoofing
   traffic to pass through.

   In the following, we expose the limitations of ACL-based ingress
   filtering and Loose uRPF for SAV at provider/peer interfaces in
   scenarios of source address spoofing from provider/peer AS.  The
   source address spoofing from provider/peer AS represents a class of
   scenario where spoofing traffic comes from a provider/peer AS and the
   spoofed source addresses belong to the customer cone which the
   spoofing traffic enters.

   Figure 6 depicts the scenario of source address spoofing from
   provider/peer AS and is used to analyze the gaps of ACL-based ingress
   filtering and Loose uRPF below.

Li, et al.                Expires 23 April 2026                [Page 14]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

                             +----------------+
               Spoofer(P1')+-+    AS 3(P3)    |
                             +-+/\----+/\+----+
                                /       \
                               /         \
                              /           \
                             / (C2P/P2P)   \
                    +----------------+      \
                    |    AS 4(P4)    |       \
                    ++/\+--+/\+--+/\++        \
       P6[AS 2, AS 1] /     |      \           \
      P1[AS 2, AS 1] /      |       \           \
           P2[AS 2] /       |        \           \
                   / (C2P)  |         \ P5[AS 5]  \ P5[AS 5]
   +----------------+       |          \           \
   |    AS 2(P2)    |       | P1[AS 1]  \           \
   +----------+/\+--+       | P6[AS 1]   \           \
                \           |             \           \
        P6[AS 1] \          |              \           \
         P1[AS 1] \         |               \           \
                   \ (C2P)  | (C2P)    (C2P) \     (C2P) \
                  +----------------+        +----------------+
                  |  AS 1(P1, P6)  |        |    AS 5(P5)    |
                  +----------------+        +----------------+
   P1' is the spoofed source prefix P1 by the spoofer which is inside of
   AS 3 or connected to AS 3 through other ASes.

       Figure 6: A scenario of source address spoofing from provider/
                                  peer AS.

   In the case of Figure 6, the spoofer which is inside of AS 3 or
   connected to AS 3 through other ASes forges the source addresses in
   P1 and sends the spoofing traffic to the destination addresses in P2.
   The arrows in Figure 6 represent the commercial relationships between
   ASes.  AS 3 acts as the provider or lateral peer of AS 4 and the
   provider for AS 5, while AS 4 serves as the provider for AS 1, AS 2,
   and AS 5.  Additionally, AS 2 is the provider for AS 1.  Suppose AS 1
   and AS 4 have deployed inter-domain SAV, while the other ASes have
   not.

   By applying ACL-based ingress filtering at the provider/peer
   interface of AS 4, the ACL rules can block any packets with spoofed
   source addresses from AS 3 in P1.  However, this approach incurs
   heavy operational overhead, as it requires network operators to
   update the ACL rules promptly based on changes in prefixes or
   topology of AS 4's customer cone.  Otherwise, it may cause improper
   block of legitimate traffic or improper permit of spoofing traffic.

Li, et al.                Expires 23 April 2026                [Page 15]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   Loose uRPF can greatly reduce the operational overhead because it
   uses the local FIB as information source, and can adapt to changes in
   the network.  However, it would improperly permit spoofed packets.
   In Figure 6, Loose uRPF is enabled at AS 4's provider/peer interface,
   while EFP-uRPF is enabled at AS 4's customer interfaces.  A spoofer
   inside AS 3 or connected to it through other ASes may send packets
   with source addresses spoofing P1 to AS 2.  As AS 3 lacks deployment
   of inter-domain SAV, the spoofing packets will reach AS 4's provider/
   peer interface.  With Loose uRPF, AS 4 cannot block them at its
   provider/peer interface facing AS 3, and thus resulting in improper
   permit.

5.  Problem Statement

   +--------+----------+-----------+----------+-------+--------+
   |Problems|    ACL   |   Strict  |  Loose   |FP-uRPF|EFP-uRPF|
   |        |          |   uRPF    |  uRPF    |       |        |
   +--------+----------+-----------+----------+-------+--------+
   |Improper|    NO    |    YES    |    NO    |      YES       |
   |Block   |(manual   | (LPP, HP) |          |    (LPP, HP)   |
   |        |operator  |           |          |                |
   |        |diligence)|           |          |                |
   +--------+----------+-----------+----------+-------+--------+
   |Improper|    NO    |NO (no SCC)|   YES    |   NO (no SCC)  |
   |Permit  |          |YES (SCC)  |  (SPP)   |   YES (SCC)    |
   +--------+----------+-----------+----------+-------+--------+
   |        |   YES    |                                       |
   |  HOO   |  (Any    |                  NO                   |
   |        |Scenarios)|                                       |
   +--------+----------+---------------------------------------+
   "HOO" refers to High Operational Overhead.
   "LPP" is a scenario with limited propagation of a prefix
   (e.g., due to NO_EXPORT Community {{noexp}}).
   "HP" is a scenario with hidden prefixes (e.g., Direct Server
   Return (DSR) {{dsrp}} anycast prefix.)
   "SPP" refers to Spoofing from Provider/Peer AS. It is a
   scenario where the spoofed traffic comes from a
   provider/peer AS and the spoofed prefix belongs to the
   customer cone of the AS that is performing SAV.
   "SCC" refers to Spoofing within a CC. It is a scenario where
   in the customer cone (CC) of the interface in consideration,
   source address spoofing occurs from within the CC,
   i.e., a prefix at one AS in the CC is spoofed from
   another AS in the CC.

Li, et al.                Expires 23 April 2026                [Page 16]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

          Figure 7: The scenarios where existing inter-domain SAV
         mechanisms may have improper block problem for legitimate
       traffic, improper permit problem for spoofing traffic, or high
                           operational overhead.

   Based on the analysis above, we conclude that existing inter-domain
   SAV mechanisms exhibit limitations in asymmetric routing scenarios,
   leading to potential issues of improper block or improper permit.
   Additionally, these mechanisms can result in high operational
   overhead, especially when network routing undergoes dynamic changes.
   Figure 7 provides a comprehensive summary of scenarios where existing
   inter-domain SAV mechanisms may encounter issues, including instances
   of improper blocking of legitimate traffic, improper permitting of
   spoofing traffic, or high operational overhead.

   For ACL-based ingress filtering, network operators need to manually
   update ACL rules to adapt to network changes.  Otherwise, they may
   cause improper block or improper permit issues.  Manual updates
   induce high operational overhead, especially in networks with
   frequent policy and route changes.

   Strict uRPF and Loose uRPF are automatic SAV mechanisms, and thus
   they do not need any manual effort to adapt to network changes.
   However, they have issues in scenarios with asymmetric routing.
   Strict uRPF may cause improper block problems when an AS is multi-
   homed and routes are not symmetrically announced to all its
   providers.  This is because the local FIB may not include the
   asymmetric routes of the legitimate packets, and Strict uRPF only
   uses the local FIB to check the source addresses and incoming
   interfaces of packets.  Loose uRPF may cause improper permit problems
   and fail to prevent source address spoofing.  This is because it is
   oblivious to the incoming interfaces of packets.

   FP-uRPF improves Strict uRPF in multi-homing scenarios.  However,
   they still have improper block issues in asymmetric routing
   scenarios.  For example, they may not handle the cases of limited
   propagation of prefixes.  These mechanisms use the local RIB to learn
   the source prefixes and their valid incoming interfaces.  But the RIB
   may not have all the prefixes with limited propagation and their
   permissible incoming interfaces.

   EFP-uRPF allows the prefixes from the same customer cone at all
   customer interfaces.  This solves the improper block problems of FP-
   uRPF in multi-homing scenarios.  However, this approach also
   compromises partial protection against spoofing from the customer
   cone.  EFP-uRPF may still have improper block problems when it does
   not learn legitimate source prefixes.  For example, hidden prefixes
   are not learned by EFP-uRPF.

Li, et al.                Expires 23 April 2026                [Page 17]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   Finally, existing inter-domain SAV mechanisms cannot work in all
   directions (i.e., interfaces) of ASes to achieve effective SAV.
   Network operators need to carefully analyze the network environment
   and choose appropriate SAV mechanism for each interface.  This leads
   to additional operational and cognitive overhead, which hinders the
   rate of adoption of inter-domain SAV.

6.  Requirements for New Inter-domain SAV Mechanisms

   This section lists the requirements which can help bridge the
   technical gaps of existing inter-domain SAV mechanisms.  These
   requirements serve as the practical guidelines that can be met, in
   part or in full, by proposing new techniques.

6.1.  Accurate Validation

   The new inter-domain SAV mechanism MUST improve the validation
   accuracy in all directions of ASes over existing inter-domain SAV
   mechanisms, while working in incremental/partial deployment and
   providing necessary security guarantee.

6.1.1.  Improving Validation Accuracy over Existing Mechanisms

   The new inter-domain SAV mechanism MUST avoid improper blocking and
   reject more spoofed traffic than existing inter-domain SAV
   mechanisms.  To achieve this, for an AS performing inter-domain SAV
   on an interface connected to a neighboring AS, it MUST permit all
   prefixes whose legitimate traffic (using them as source addresses)
   can reach that interface, while blocking all other prefixes that
   cannot.  This general principle applies to customer, lateral peer,
   and provider interfaces.  Multiple sources of SAV-related
   information, such as ROA and ASPA objects, BGP Update data, SAV-
   specific information, and management information can be leveraged to
   meet this requirement.

Li, et al.                Expires 23 April 2026                [Page 18]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

                         +-----------------+
                         |       AS 4      |
                         +------+/\+-------+
                 Spoofing traffic | Legitimate traffic
                 with SA in P1'   | with SA in P1
                                  | (C2P)
                         +-----------------+
                         |       AS 3      |
                         +--+/\+-----+/\+--+
           Spoofing traffic  /         \ Legitimate traffic
           with SA in P1'   /           \ with SA in P1
                           / (C2P) (C2P) \
                +----------------+   +----------------+
   Spoofer(P1')-+    AS 2(P2)    |   |    AS 1(P1)    |
                +----------------+   +----------------+
   P1' is the spoofed source prefix P1 by the spoofer which is inside of
   AS 2 or connected to AS 2 through other ASes.
   AS 4 performs SAV at its interface facing AS 3.
   The legitimate traffic with SA in P1 arrives at AS 4 along the path
   AS 1->AS 3->AS 4.
   The spoofing traffic with SA in P1' arrives at AS 4 along the path
   AS 2->AS 3->AS 4.

      Figure 8: An example where both spoofing and legitimate traffic
                      arrive from the same direction.

   The path taken by the traffic with spoofed source address (i.e.,
   spoofed traffic) may overlap with a path for the legitimate traffic.
   Such scenarios could result in improper permit of the spoofed traffic
   at the AS doing SAV unless an AS located at or prior to the merging
   point of the overlap is also performing inter-domain SAV.  As
   illustrated in Figure 8, both spoofed and legitimate traffic traverse
   the same link between AS 3 and AS 4.  In this case, SAV filtering at
   AS 4's interface facing AS 3 cannot differentiate between the two.
   The spoofed traffic in such scenarios is incrementally mitigated
   (i.e., blocked) with the wider deployment of SAV.  For example, AS 3
   can deploy SAV on its interfaces facing AS 1 and AS 2 to facilitate
   blocking of the spoofed traffic while admitting and propagating the
   legitimate traffic.

6.1.2.  Working in Incremental/Partial Deployment

   The new inter-domain SAV mechanism MUST NOT assume pervasive adoption
   (including the adoption of both SAV and SAV-related information) and
   SHOULD benefit early adopters by providing effective protection from
   spoofing of source addresses even when it is partially deployed in
   the Internet.  Not all AS border routers can support the new SAV
   mechanism at once, due to various constraints such as capabilities,

Li, et al.                Expires 23 April 2026                [Page 19]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   versions, or vendors.  The new SAV mechanism SHOULD NOT be less
   effective than existing mechanisms in its capability of protection
   from source address spoofing for any type of peering interface
   (customer, lateral peer, and provider) even under partial deployment.

6.1.3.  Providing Necessary Security Guarantee

   The new inter-domain SAV mechanism SHOULD secure the communicated
   SAV-related information between ASes.  It SHOULD prevent malicious
   ASes from generating forged information or detect and filter the
   forged information from malicious ASes.

6.2.  Automatic Update

   The new inter-domain SAV mechanism SHOULD update SAV rules and detect
   the changes of SAV-specific information automatically while
   guaranteeing convergence.

6.2.1.  Reducing Operational Overhead

   The new inter-domain SAV mechanism MUST be able to adapt to dynamic
   networks and asymmetric routing scenarios automatically, instead of
   relying on manual update.  At least, it MUST have less operational
   overhead than ACL-based ingress filtering.

6.2.2.  Guaranteeing Convergence

   The new inter-domain SAV mechanism SHOULD promptly detect the network
   changes and finish the convergence process quickly.  It is essential
   that the new inter-domain SAV mechanism converges towards accurate
   SAV rules in a proper manner, effectively reducing improper block and
   improper permit throughout the whole convergence process.

7.  Inter-domain SAV Scope

   The new inter-domain SAV mechanisms should work in the same scenarios
   as existing ones.  Generally, it includes all IP-encapsulated
   scenarios:

   *  Native IP forwarding: This includes both global routing table
      forwarding and CE site forwarding of VPN.

   *  IP-encapsulated Tunnel (IPsec, GRE, SRv6, etc.): In this scenario,
      we focus on the validation of the outer layer IP address.

   *  Both IPv4 and IPv6 addresses.

   Scope does not include:

Li, et al.                Expires 23 April 2026                [Page 20]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   *  Non-IP packets: This includes MPLS label-based forwarding and
      other non-IP-based forwarding.

   In addition, the new inter-domain SAV mechanisms should not modify
   data plane packets.  Existing architectures or protocols or
   mechanisms can be inherited by the new SAV mechanism to achieve
   better SAV effectiveness.

8.  Security Considerations

   SAV rules can be generated based on route information (FIB/RIB) or
   non-route information.  If the information is poisoned by attackers,
   the SAV rules will be false.  Legitimate packets may be dropped
   improperly or malicious traffic with spoofed source addresses may be
   permitted improperly.  Route security should be considered by routing
   protocols.  Non-route information, such as RPKI ASPA objects, should
   also be protected by corresponding mechanisms or infrastructure.  If
   SAV mechanisms or protocols require exchanging specific information
   between ASes, some considerations on the avoidance of message
   alteration or message injection are needed to propose.

9.  IANA Considerations

   This document does not request any IANA allocations.

10.  Contributors

   Nan Geng
   Huawei
   Beijing, China
   Email: gengnan@huawei.com

11.  References

11.1.  Normative References

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/rfc/rfc8174>.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/rfc/rfc2119>.

   [RFC3704]  Baker, F. and P. Savola, "Ingress Filtering for Multihomed
              Networks", BCP 84, RFC 3704, DOI 10.17487/RFC3704, March
              2004, <https://www.rfc-editor.org/rfc/rfc3704>.

Li, et al.                Expires 23 April 2026                [Page 21]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   [RFC8704]  Sriram, K., Montgomery, D., and J. Haas, "Enhanced
              Feasible-Path Unicast Reverse Path Forwarding", BCP 84,
              RFC 8704, DOI 10.17487/RFC8704, February 2020,
              <https://www.rfc-editor.org/rfc/rfc8704>.

   [RFC2827]  Ferguson, P. and D. Senie, "Network Ingress Filtering:
              Defeating Denial of Service Attacks which employ IP Source
              Address Spoofing", BCP 38, RFC 2827, DOI 10.17487/RFC2827,
              May 2000, <https://www.rfc-editor.org/rfc/rfc2827>.

   [RFC4364]  Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private
              Networks (VPNs)", RFC 4364, DOI 10.17487/RFC4364, February
              2006, <https://www.rfc-editor.org/rfc/rfc4364>.

   [RFC6811]  Mohapatra, P., Scudder, J., Ward, D., Bush, R., and R.
              Austein, "BGP Prefix Origin Validation", RFC 6811,
              DOI 10.17487/RFC6811, January 2013,
              <https://www.rfc-editor.org/rfc/rfc6811>.

   [RFC4786]  Abley, J. and K. Lindqvist, "Operation of Anycast
              Services", BCP 126, RFC 4786, DOI 10.17487/RFC4786,
              December 2006, <https://www.rfc-editor.org/rfc/rfc4786>.

   [RFC7094]  McPherson, D., Oran, D., Thaler, D., and E. Osterweil,
              "Architectural Considerations of IP Anycast", RFC 7094,
              DOI 10.17487/RFC7094, January 2014,
              <https://www.rfc-editor.org/rfc/rfc7094>.

11.2.  Informative References

   [RFC5210]  Wu, J., Bi, J., Li, X., Ren, G., Xu, K., and M. Williams,
              "A Source Address Validation Architecture (SAVA) Testbed
              and Deployment Experience", RFC 5210,
              DOI 10.17487/RFC5210, June 2008,
              <https://www.rfc-editor.org/rfc/rfc5210>.

   [RFC7908]  Sriram, K., Montgomery, D., McPherson, D., Osterweil, E.,
              and B. Dickson, "Problem Definition and Classification of
              BGP Route Leaks", RFC 7908, DOI 10.17487/RFC7908, June
              2016, <https://www.rfc-editor.org/rfc/rfc7908>.

   [RFC9234]  Azimov, A., Bogomazov, E., Bush, R., Patel, K., and K.
              Sriram, "Route Leak Prevention and Detection Using Roles
              in UPDATE and OPEN Messages", RFC 9234,
              DOI 10.17487/RFC9234, May 2022,
              <https://www.rfc-editor.org/rfc/rfc9234>.

Li, et al.                Expires 23 April 2026                [Page 22]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   [manrs]    MANRS, "MANRS Implementation Guide", 2023,
              <https://www.manrs.org/netops/guide/antispoofing/>.

   [isoc]     Internet Society, "Addressing the challenge of IP
              spoofing", 2015,
              <https://www.internetsociety.org/resources/doc/2015/
              addressing-the-challenge-of-ip-spoofing/>.

   [nist]     NIST, "Border Gateway Protocol Security and Resilience",
              2025, <https://doi.org/10.6028/NIST.SP.800-189r1.ipd>.

   [urpf]     Cisco Systems, Inc., "Unicast Reverse Path Forwarding
              Enhancements for the Internet Service Provider-Internet
              Service Provider Network Edge", 2005,
              <https://www.cisco.com/c/dam/en_us/about/security/
              intelligence/urpf.pdf>.

   [bar-sav]  NIST, Akamai, "Source Address Validation Using BGP
              UPDATEs, ASPA, and ROA (BAR-SAV)", 2024,
              <https://datatracker.ietf.org/doc/draft-ietf-sidrops-bar-
              sav/>.

Acknowledgements

   Many thanks to Jared Mauch, Barry Greene, Fang Gao, Anthony Somerset,
   Yuanyuan Zhang, Igor Lubashev, Alvaro Retana, Joel Halpern, Aijun
   Wang, Michael Richardson, Li Chen, Gert Doering, Mingxing Liu, John
   O'Brien, and Roland Dobbins, for their valuable comments on this
   document.  Apologies to any others whose names the authors may have
   missed mentioning.

Authors' Addresses

   Dan Li
   Tsinghua University
   Beijing
   China
   Email: tolidan@tsinghua.edu.cn

   Lancheng Qin
   Zhongguancun Laboratory
   Beijing
   China
   Email: qinlc@zgclab.edu.cn

Li, et al.                Expires 23 April 2026                [Page 23]
Internet-Draft    Inter-domain SAVNET Problem Statement     October 2025

   Libin Liu
   Zhongguancun Laboratory
   Beijing
   China
   Email: liulb@zgclab.edu.cn

   Mingqing Huang
   Huawei
   Beijing
   China
   Email: huangmingqing@huawei.com

   Kotikalapudi Sriram
   USA National Institute of Standards and Technology
   Gaithersburg, MD
   United States of America
   Email: sriram.ietf@gmail.com

Li, et al.                Expires 23 April 2026                [Page 24]