Inter-domain Source Address Validation (SAVNET) Architecture

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 5 September 2024.¶

Copyright Notice

Copyright (c) 2024 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.¶

Table of Contents

1. Introduction

Attacks based on source IP address spoofing, such as reflective DDoS and flooding attacks, continue to present significant challenges to Internet security. Mitigating these attacks in inter-domain networks requires effective source address validation (SAV). While BCP84 [RFC3704] [RFC8704] offers some SAV solutions, such as ACL-based ingress filtering and uRPF-based mechanisms, existing inter-domain SAV mechanisms have limitations in terms of validation accuracy and operational overhead in different scenarios [inter-domain-ps].¶

There are various existing general information from different sources including RPKI ROA objects and ASPA objects, RIB, FIB, and Internet Routing Registry (IRR) data, which can be used for inter-domain SAV. Generating SAV rules based on general information, however, cannot well satisfy the requirements for new inter-domain SAV mechanisms proposed in [inter-domain-ps]. As analyzed in Section 5, general information from RPKI ROA objects and ASPA objects can be used to infer the prefixes and their permissible incoming directions yet cannot be updated in a timely manner to adapt to the prefix or route changes, and the local routing information, which represents the general information from RIB or FIB, cannot deal with the asymmetric routing scenarios and may lead to improper blocks or improper permits, while IRR data do not update in a timely manner either and are not always accurate.¶

Consequently, to address these issues, the inter-domain SAVNET architecture focuses on providing a comprehensive framework and guidelines for the design and implementation of new inter-domain SAV mechanisms. Inter-domain SAVNET architecture proposes SAV-specific information and uses it to generate SAV rules. SAV-specific information consists of prefixes and their corresponding legitimate incoming direction to enter an AS. Inter-domain SAVNET architecture can use it to generate more accurate SAV rules. In order to gather the SAV-specific information, a SAV-specific information communication mechanism would be developed for origination, processing, propagation, and termination of the messages which carry the SAV-specific information, and it can be implemented by a new protocol or extending an existing protocol. When the prefixes or routes change, it can update the SAV-specific information automatically in a timely manner. Also, the inter-domain SAVNET architecture will communicate the SAV-specific information over a secure connection between authenticated ASes.¶

Moreover, during the incremental/partial deployment period of the SAV-specific information, the inter-domain SAVNET architecture can leverage the general information to generate SAV rules, if the SAV-specific information of an AS is unavailable. Multiple information sources may exist concurrently, to determine the one used for generating SAV rules, the inter-domain SAVNET architecture assigns priorities to the SAV-specific information and different general information and generates SAV rules using the SAV-related information with the highest-priority. SAV-specific information has the highest priority and the priorities of RPKI ROA objects and ASPA objects, RIB, FIB, and IRR data decrease in turn.¶

+-----------+
| AS 1 (P1) #
+-----------+ \
               \           Spoofed Packets
             +-+#+-------+ with Source Addresses in P1 +-----------+
             |    AS 2   #-----------------------------#   AS 4    |
             +-+#+-------+                             +-----------+
               /
+-----------+ /
|   AS 3    #
+-----------+
AS 4 sends spoofed packets with source addresses in P1 to AS 3
through AS 2.
If AS 1 and AS 2 deploy inter-domain SAV, the spoofed packets
can be blocked at AS 2.

The inter-domain SAVNET architecture provides the incentive to deploy inter-domain SAV for operators. Figure 1 illustrates this using an example. P1 is the source prefix of AS 1, and AS 4 sends spoofing packets with P1 as source addresses to AS 3 through AS 2. Assume AS 4 does not deploy intra-domain SAV, these spoofing packets cannot be blocked by AS 4. Although AS 1 can deploy intra-domain SAV to block incoming packets which spoof the addresses of AS 1, these spoofing traffic from AS 4 to AS 3 do not go through AS 1, so they cannot be blocked by AS 1. Inter-domain SAVNET architecture can help in this scenario. If AS 1 and AS 2 deploy inter-domain SAVNET architecture, AS 2 knows that the packets with P1 as source addresses should come from AS 1, and the spoofing packets can thus be blocked by AS 2 since they come from the incorrect direction. Specifically, by proposing SAV-specific information and using it to generate SAV rules, the inter-domain SAVNET architecture gives more deployment incentive compared to existing inter-domain SAV mechanisms, which will be analyzed in Section 8.¶

In addition, this document primarily proposes a high-level architecture for describing the communication flow of SAV-specific information and general information, guiding how to utilize the SAV-specific information and general information for generating SAV rules and deploy an inter-domain SAV mechanism between ASes. This document does not specify protocol extensions or implementations. Its purpose is to provide a conceptual framework and guidance for the design and development of inter-domain SAV mechanisms, allowing implementers to adapt and implement the architecture based on their specific requirements and network environments.¶

2. Terminology

SAV Rule:

The rule that indicates the validity of a specific source IP address or source IP prefix.¶

SAV Table:

The table or data structure that implements the SAV rules and is used for performing source address validation on the data plane.¶

SAV-specific Information:

The information that is specialized for SAV rule generation, includes the source prefixes and their legitimate incoming directions to enter an AS, and is gathered by the communication between ASes with the SAV-specific information communication mechanism.¶

SAV-specific Information Communication Mechanism:

The mechanism that is used to communicate SAV-specific information between ASes and can be implemented by a new protocol or an extension to an existing protocol.¶

Local Routing Information:

The information that is stored in ASBR's local RIB or FIB and can be used to generate SAV rules in addition to the routing purpose.¶

General Information:

The information that is not specialized for SAV but can be utilized to generate SAV rules, and is initially utilized for other purposes. Currently, the general information consists of the information from RPKI ROA objects and ASPA objects, local routing information, and the one from IRR data.¶

SAV-related Information:

The information that can be used to generate SAV rules and includes SAV-specific information and general information.¶

SAVNET Agent:

The agent within a SAVNET-adopting AS that is responsible for gathering SAV-related information and utilizing it to generate SAV rules.¶

SAV Information Base:

SAV information base is a table or data structure for storing SAV-related information collected from different SAV information sources and is a component within SAVNET agent.¶

SAV Information Base Manager:

SAV information base manager maitains the SAV-related information in the SAV information base and uses it to generate SAV rule accordingly, and is a component within SAVNET agent.¶

Improper Block:

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. Design Goals

The inter-domain SAVNET architecture aims to improve SAV accuracy and facilitate partial deployment with low operational overhead, while guaranteeing convergence and providing security guarantees to the communicated information, which corresponds to the requirements for new inter-domain SAV mechanisms proposed in the inter-domain SAVNET architecture draft [inter-domain-ps]. The overall goal can be broken down into the following aspects:¶

• G1: The inter-domain SAVNET architecture should learn the real paths of source prefixes to any destination prefixes or permissible paths that can cover their real paths, and generate accurate SAV rules automatically based on the learned information to avoid improper blocks and reduce improper permits as much as possible.¶

• G2: The inter-domain SAVNET architecture should provide sufficient protection for the source prefixes of ASes that deploy it, even if only a portion of the Internet does the deployment.¶

• G3: The inter-domain SAVNET architecture should adapt to dynamic networks and asymmetric routing scenarios automatically.¶

• G4: The inter-domain SAVNET architecture should promptly detect the network changes and launch the convergence process in a timely manner, while reducing improper blocks and improper permits during the convergence process.¶

• G5: The inter-domain SAVNET architecture should provide security guarantees for the communicated SAV-specific information.¶

Other design goals, such as low operational overhead and easy implementation, are also very important and should be considered in specific protocols or protocol extensions.¶

4. Inter-domain SAVNET Architecture Overview

 +~~~~~~~~~~~~~~~~~~~~~~~~+       +--------------------+
 |RPKI Cache Server/IRR DB|       | AS X's Provider AS |
 +~~~~~~~~~~~~~~~~~~~~~~~~+       +------------+/\+/\+-+
   ROA & ASPA |                 BGP /            |  |
Obj./IRR Data |            Message /             |  |
              |                   /              |  | BGP
+-----------+\/+----------------+\/+-+           |  | Message
|                AS X                |  BGP  +--------------+
| +--------------------------------+ |<------|AS X's Lateral|
| |          SAVNET Agent          | |Message|   Peer  AS   |
| +--------------------------------+ |       +--------------+
+-----+/\+/\+----------------+/\+/\+-+
        |  |                   |  |
    BGP |  | SAV-specific      |  |
Message |  | Message           |  |
+------------------+           |  |
|AS X's Customer AS|           |  |
+-------+/\+-------+           |  |
          \                    |  |
       BGP \  SAV-specific     |  | BGP
    Message \ Message          |  | Message
    +------------------------------------+
    |          AS X's Customer AS        |
    | +--------------------------------+ |
    | |         SAVNET Agent           | |
    | +--------------------------------+ |
    +------------------------------------+
AS X and one of its customer ASes have deployed SAVNET agent
and can exchange SAV-specific information with each other.

Figure 2 provides an overview of the inter-domain SAVNET architecture, showcasing an AS topology and the flow of SAV-related information among ASes. The topology captures the full spectrum of AS relationships in the Internet, displaying all peer ASes of AS X including customers, lateral peers, and providers and the existence of multiple physical links between ASes. Arrows in the figure indicate the direction of the corresponding SAV-related information from its source to AS X, such as gathering RPKI ROA objects and ASPA objects from RPKI cache server. The inter-domain SAVNET architecture conveys the SAV-related information through various mediums such as SAV-specific messages, BGP messages, RTR messages, and FTP messages. Based on the SAV-related information, AS X generates SAV rules. It is also worth noting that the inter-domain SAVNET architecture discusses AS-level inter-domain SAV.¶

Figure 2 uses AS X as the representative to illustrate that what SAV-related information the SAVNET agent within AS X will collect and where the information is from. AS X has deployed SAVNET agent and can generate SAV rules to perform inter-domain SAV by consolidating the SAV-related information. It can obtain SAV-specific information from its customer AS which deploys SAVNET agent and local routing information originating from the BGP update messages of its neighbor ASes. Also, AS X can obtain RPKI ROA objects and ASPA objects from RPKI cache server and IRR data from IRR database.¶

The inter-domain SAVNET architecture proposes SAV-specific information, which is more accurate and trustworthy than existing general information, and can update in a timely manner. SAV-specific information consists of prefixes and their legitimate incoming directions. The SAVNET agent communicates SAV-specific information between ASes via SAV-specific messages, when prefixes or routes change, it can launch SAV-specific messages timely to update SAV-specific information. Additionally, when SAVNET agent receives SAV-specific messages, it will validate whether the SAV-specific connections for communicating SAV-specific messages are authentic connections from authenticated ASes. Therefore, when SAV-specific information of an AS is available, SAVNET agent will use it to generate SAV rules.¶

Furthermore, if the SAV-specific information is needed to communicate between ASes, a new SAV-specific information communication mechanism would be developed to exchange the SAV-specific messages between ASes which carry the SAV-specific information. It should define the data structure or format for communicating the SAV-specific information and the operations and timing for originating, processing, propagating, and terminating the SAV-specific messages. Also, it can be implemented by a new protocol or extending an existing protocol.¶

The SAVNET agent should launch SAV-specific messages to adapt to the route changes in a timely manner. The SAV-specific information communication mechanism should handle route changes carefully to avoid improper blocks. The reasons for leading to improper blocks may include late detection of route changes, delayed message transmission, or packet losses. During the convergence process of the SAV-specific information communication mechanism, the inter-domain SAVNET architecture can use the information from RPKI ROA objects and ASPA objects to generate SAV rules until the convergence process is finished, since these information includes topological information and is more stable, and can thus avoid improper blocks. However, the detailed design of the SAV-specific information communication mechanism for dealing with route changes is outside the scope of this document.¶

In the incremental/partial deployment stage of the inter-domain SAVNET architecture, when the SAV-specific information of some ASes is unavailable, SAVNET agent can leverage general information to generate SAV rules. If all these general information is available, it is recommended to use RPKI ROA objects and ASPA objects to generate SAV rules. Since compared to the local routing information and IRR data, they can provide authoritative prefixes and topological information and have less improper blocks. The systematic recommendations for the utilizations of SAV-related information and the corresponding rationale will be illustrated in Section 6.¶

Regarding the security concerns, the inter-domain SAVNET architecture shares the similar security threats with BGP and can leverage existing BGP security mechanisms to enhance both session and content security.¶

+-----------------------------------------------------------+
|                         Other ASes                        |
+-----------------------------------------------------------+
                                           | SAV-specific
                                           | Messages
+-----------------------------------------------------------+
|                           AS X           |                |
| +-------------------------------------------------------+ |
| |                      SAVNET Agent      |              | |
| |                                       \/              | |
| | +---------------------+  +--------------------------+ | |
| | | General Information |  | SAV-specific Information | | |
| | +---------------------+  +--------------------------+ | |
| |            |                           |              | |
| |           \/                          \/              | |
| | +---------------------------------------------------+ | |
| | | +-----------------------------------------------+ | | |
| | | |              SAV Information Base             | | | |
| | | +-----------------------------------------------+ | | |
| | |            SAV Information Base Manager           | | |
| | +---------------------------------------------------+ | |
| |                          |SAV Rules                   | |
| +-------------------------------------------------------+ |
|                            |                              |
|                           \/                              |
| +-------------------------------------------------------+ |
| |                      SAV Table                        | |
| +-------------------------------------------------------+ |
+-----------------------------------------------------------+

Figure 3 displays the SAVNET agent and SAV table within AS X. The SAVNET agent can obtain the SAV-specific information and general information from various SAV information sources including SAV-specific messages from other ASes, RPKI cache server, and RIB or FIB as long as they are available. The SAV information base (SIB) within the SAVNET agent can store the SAV-specific information and general information and is maintained by the SIB manager. And SIB manager generates SAV rules based on the SIB and fills out the SAV table on the data plane. Moreover, the SIB can be managed by network operators using various methods such as YANG [RFC6020], Command-Line Interface (CLI), remote triggered black hole (RTBH) [RFC5635], and Flowspec [RFC8955]. The detailed collection methods of the SAV-related information depend on the deployment and implementation of the inter-domain SAV mechanisms and are out of scope for this document.¶

In the data plane, the packets coming from other ASes will be validated by the SAV table and only the packets which are permitted by the SAV table will be forwarded to the next hop. To achieve this, the router looks up each packet's source address in its local SAV table and gets one of three validity states: "Valid", "Invalid" or "Unknown". "Valid" means that there is a source prefix in SAV table covering the source address of the packet and the valid incoming interfaces covering the actual incoming interface of the packet. According to the SAV principle, "Valid" packets will be forwarded. "Invalid" means there is a source prefix in SAV table covering the source address, but the incoming interface of the packet does not match any valid incoming interface so that such packets will be dropped. "Unknown" means there is no source prefix in SAV table covering the source address. The packet with "unknown" addresses can be dropped or permitted, which depends on the choice of operators. The structure and detailed usage of SAV table can refer to [sav-table].¶

5. SAV-related Information

SAV-related information represents the information that can be used for SAV and consists of RPKI ROA objects and ASPA objects, local routing information, IRR data, and SAV-specific information. In the inter-domain SAVNET architecture, RPKI ROA objects and ASPA objects, local routing information, and IRR data are categorized into general information. In the future, if a new information source is created and can be used for SAV, but is not originally and specially used for SAV, its information can be categorized into general information. In other words, general information can also be considered as dual-use information.¶

+-------------------------+-----------+----------+---------------+
| SAV-related Information |  Accurate |Real-time |Trustworthiness|
|                         |    SAV    | Update   |               |
+-----------+-------------+-----------+----------+---------------+
|           |RPKI ROA Obj.|           |    NO    |      YES      |
|           | & ASPA Obj. |           |          |               |
|           +-------------+  Improper +----------+---------------+
|General    |Local Routing|   Block   |    YES   |       NO      |
|Information| Information |     &     |          |               |
|           +-------------+  Improper +----------+---------------+
|           |  IRR Data   |   Permit  |    NO    |       NO      |
|           |             |           |          |               |
+-----------+-------------+-----------+----------+---------------+
|                         |Functioning|          |               |
|SAV-specific Information |    as     |    YES   |       YES     |
|                         | Expected  |          |               |
+-------------------------+-----------+----------+---------------+

6. SAV Information Base

+---------------------------------------------------+----------+
|              SAV Information Sources              |Priorities|
+---------------------------------------------------+----------+
|              SAV-specific Information             |     1    |
+---------------------+-----------------------------+----------+
|                     | RPKI ROA Obj. and ASPA Obj. |     2    |
|                     +-----------------------------+----------+
|                     |             RIB             |     3    |
| General Information +-----------------------------+----------+
|                     |             FIB             |     4    |
|                     +-----------------------------+----------+
|                     |           IRR Data          |     5    |
+---------------------+-----------------------------+----------+
Priority ranking from 1 to 5 represents high to low priority.

The SIB is managed by the SIB manager, which can consolidate SAV-related information from different sources. Figure 5 presents the priority ranking for the SAV-specific information and general information. Priority ranking from 1 to 5 represents high to low priority. Inter-domain SAVNET architecture uses SAV-related information from different sources based on their priorities. Once the SAV-specific information for a prefix is available within the SIB, the inter-domain SAVNET generates SAV rules based on SAV-specific information; otherwise, the inter-domain SAVNET generates SAV rules based on general information. The inter-domain SAVNET architecture assigns priorities to the information from different SAV information sources, and always generates the SAV rules using the information with the highest priority from all the available information.¶

The priority ranking recommendation for different SAV information sources in Figure 5 is based on the accuracy, timeliness, trustness of the information from these sources. These properties determine that whether the requirements for new inter-domain SAV mechanisms proposed in [inter-domain-ps] can be well satisfied. SAV-specific information has higher priority than the general information, since it is specifically designed to carry more accurate SAV information, and can be updated in a timely manner to adapt to the prefix or route changes. The general information from RPKI ROA objects and ASPA objects, RIB, FIB, IRR data has different priorities, ranking 2, 3, 4, and 5, respectively. RPKI ROA objects and ASPA objects have higher priority than the one from RIB, FIB, and IRR data, this is because they can provide authoritative prefixes and topology information, which can be used to generate more accurate SAV rules. Also, they are more stable and can be used to reduce the risk of improper blocks during the convergence process of the network. Although the information source for RIB and FIB is the same, the RIB consists of more backup path information than the FIB, which can reduce improper blocks. IRR data have the lowest priority compared to others, since they are usually updated in a slower manner than the real network changes and not always correct.¶

                        +----------------+
                        |    AS 3(P3)    |
                        +-+/\-----+/\+/\++
                           /        \  \
                 P3[AS 3] /          \  \ P3[AS 3]
                         /            \  \
                        / (C2P)        \  \
               +----------------+       \  \
               |    AS 4(P4)    |        \  \
               ++/\+/\+/\+/\+/\++         \  \
  P6[AS 1, AS 2] /  /  |  |    \           \  \
       P2[AS 2] /  /   |  |     \           \  \
               /  /    |  |      \           \  \
              /  /     |  |       \ P5[AS 5]  \  \ P5[AS 5]
             /  /      |  |        \           \  \
            /  /(C2P)  |  |         \           \  \
+----------------+     |  |          \           \  \
|    AS 2(P2)    |     |  | P1[AS 1]  \           \  \
+--------+/\+----+     |  | P6[AS 1]   \           \  \
  P6[AS 1] \           |  | NO_EXPORT   \           \  \
   P1[AS 1] \          |  |              \           \  \
   NO_EXPORT \         |  |               \           \  \
              \ (C2P)  |  | (C2P/P2P) (C2P)\     (C2P) \  \
           +----------------+              +----------------+
           |  AS 1(P1, P6)  |              |    AS 5(P5)    |
           +----------------+              +----------------+
Both AS 1 and AS 4 deploy the inter-domain SAVNET architecture
and can exchange the SAV-specific information with each other,
while other ASes do not deploy it.

+-----+------+------------------+--------+------------------------+
|Index|Prefix|Incoming Direction|Relation| SAV Information Source |
+-----+------+------------------+--------+------------------------+
|  0  |  P1  |       AS 2       |Customer|SAV-specific Information|
+-----+------+------------------+--------+------------------------+
|  1  |  P1  |       AS 1       |Customer|  General Information   |
+-----+------+------------------+--------+------------------------+
|  2  |  P2  |       AS 2       |Customer|  General Information   |
+-----+------+------------------+--------+------------------------+
|  3  |  P3  |       AS 3       |Provider|  General Information   |
+-----+------+------------------+--------+------------------------+
|  4  |  P5  |       AS 3       |Provider|  General Information   |
+-----+------+------------------+--------+------------------------+
|  5  |  P5  |       AS 5       |Customer|  General Information   |
+-----+------+------------------+--------+------------------------+
|  6  |  P6  |       AS 2       |Customer|  General Information   |
|     |      |                  |        |SAV-specific Information|
+-----+------+------------------+--------+------------------------+
|  7  |  P6  |       AS 1       |Customer|  General Information   |
+-----+------+------------------+--------+------------------------+

We use the examples shown in Figure 6 and Figure 7 to introduce SIB and illustrate how to generate SAV rules based on the SIB. Figure 7 depicts an example of the SIB established in AS 4 displayed in Figure 6. Each row of the SIB contains an index, prefix, incoming direction of the prefix, reltation between ASes, and the corresponding sources of the information. The incoming direction consists of customer, provider, and peer. For example, in Figure 7, the row with index 0 indicates the incoming direction of P1 is AS 2 and the information source is SAV-specific information. Note that the same SAV-related information may have multiple sources and the SIB records them all, such as the row indexed 6. Moreover, SIB should be carefully implemented in the specific protocol or protocol extensions to avoid becoming a heavy burden of the router, and the similar optimization approaches used for the RIB may be applied.¶

Recall that inter-domain SAVNET architecture generates SAV rules based on the SAV-related information in the SIB and their priorities. In addition, in the case of an AS's interfaces facing provider or lateral peer ASes where loose SAV rules are applicable, the inter-domain SAVNET architecture recommends to use blocklist at such directions to only block the prefixes that are sure not to come at these directions, while in the case of an AS's interfaces facing customer ASes that necessitate stricter SAV rules, the inter-domain SAVNET architecture recommends to use allowlist to only permit the prefixes that are allowed to come at these directions.¶

Based on the above rules, taking the SIB in Figure 7 as an example to illustrate how the inter-domain SAVNET generates rules, AS 4 can conduct SAV as follows: SAV at the interfaces facing AS 3 blocks P1, P2, and P6 according to the rows indexed 0, 2, and 6 in the SIB, SAV at the interfaces facing AS 2 permits P1, P2, and P6 according to the rows indexed 0, 2, and 6 in the SIB, SAV at the interfaces facing AS 1 does not permit any prefixes according to the row indexed 0, 1, 6, and 7 in the SIB, and SAV at the interfaces facing AS 5 permits P5 according to the row indexed 5 in the SIB.¶

7. SAVNET Communication Mechanism

+------+  SAV-specific Information
|      |  Communication Mechanism  +-----------------------------+
|      |<==========================|  SAVNET Agent in other ASes |
|      |                           +-----------------------------+
|      |                         +---------------------------------+
|      |                         | +-----------------------------+ |
|      |                         | | RPKI ROA Obj. and ASPA Obj. | |
|      |                         | +-----------------------------+ |
|      |                         | +-----------------------------+ |
|      |  General Information    | |             RIB             | |
|SAVNET| Communication Mechanism | +-----------------------------+ |
|Agent |<------------------------| +-----------------------------+ |
|      |                         | |             FIB             | |
|      |                         | +-----------------------------+ |
|      |                         | +-----------------------------+ |
|      |                         | |         IRR Database        | |
|      |                         | +-----------------------------+ |
|      |                         +---------------------------------+
|      |  Management Mechanism   +-----------------------------+
|      |<------------------------|      Network Operators      |
|      |                         +-----------------------------+
+------+

SAV-specific information relies on the communication between SAVNET agents and general information can be from RPKI ROA objects and ASPA objects, RIB, FIB, and IRR data. Therefore, as illustrated in Figure 8, the SAVNET agent needs to receive the SAV-related information from these SAV information sources. SAVNET agent also needs to accept the configurations from network operators for the management operations. Gathering these types of information relies on the SAVNET communication mechanism, which includes SAV-specific information communication mechanism, general information communication mechanism, and management mechanism.¶

+------------------+                        +------------------+
|   AS 1 (P1, P6)  | SAV-specific Messages  |     AS X (P4)    |
| +-------------+  | (P1, AS 2), (P6, AS 2) |  +-------------+ |
| |    SAVNET   |--|------------------------|->|    SAVNET   | |
| |    Agent    |<-|------------------------|--|    Agent    | |
| +-------------+  |      (P4, AS X)        |  +-------------+ |
+------------------+  SAV-specific Messages +------------------+

8. Use Cases

This section utilizes the sample use cases to showcase that the inter-domain SAVNET architecture can improve the validation accuracy in the scenarios of limited propagation of prefixes, hidden prefixes, reflection attacks, and direct attacks, compared to existing SAV mechanisms, which are also utilized for the gap analysis of existing inter-domain SAV mechanisms in [inter-domain-ps]. In the following, these use cases are discussed for SAV at customer interfaces and SAV at provider/peer interfaces, respectively.¶

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

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

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

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

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

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

9. Partial/Incremental Deployment Considerations

The inter-domain SAVNET architecture MUST ensure support for partial/incremental deployment as it is not feasible to deploy it simultaneously in all ASes. The partial/incremental deployment of the inter-domain SAVNET architecture consists of different aspects, which include the partial/incremental deployment of the architecture and the partial/incremental deployment of the information sources.¶

Within the architecture, the general information like the prefixes and topological information from RPKI ROA Objects and ASPA Objects and the routing information from the RIB can be obtained locally when the corresponding sources are available. Even when both SAV-specific Information and the information from RPKI ROA Objects and ASPA Objects are not available, the routing information from the RIB can be used to generate SAV rules.¶

Furthermore, it is not mandatory for all ASes to deploy SAVNET agents for SAV-specific Information. Instead, a SAVNET agent should be able to effortlessly establish a logical neighboring relationship with another AS that has deployed a SAVNET agent. The connections for communicating SAV-specific Information can be achieved by manual configurations set by operators or an automatic neighbor discovery mechanism. This flexibility enables the architecture to accommodate varying degrees of deployment, promoting interoperability and collaboration among participating ASes. During the partial/incremental deployment of SAVNET agent, the SAV-specific Information for the ASes which do not deploy SAVNET agent can not be obtained. To protect the prefixes of these ASes, inter-domain SAVNET architecture can use the SAV-related information from the general information in the SIB to generate SAV rules. At least, the routing information from the RIB can be always available in the SIB.¶

As more ASes adopt the inter-domain SAVNET architecture, the "deployed area" expands, thereby increasing the collective defense capability against source address spoofing. Furthermore, if multiple "deployed areas" can be logically interconnected across "non-deployed areas", these interconnected "deployed areas" can form a logical alliance, providing enhanced protection against address spoofing. Especially, along with more ASes deploy SAVNET agent and support the communication of SAV-specific information, the generated SAV rules of the inter-domain SAVNET architecture to protect these ASes will become more accurate, as well as enhancing the protection capability against source address spoofing for the inter-domain SAVNET architecture.¶

In addition, releasing the SAV functions of the inter-domain SAVNET architecture incrementally is one potential way to reduce the deployment risks and can be considered in its deployment by network operators:¶

• First, the inter-domain SAVNET can only do the measurement in the data plane and do not take any other actions. Based on the measurement data, the operators can evaluate the effect of the inter-domain SAVNET on the legitimate traffic, including validation accuracy and forwarding performance, as well as the operational overhead.¶

• Second, the inter-domain SAVNET can open the function to limit the rate of the traffic that is justified as spoofing traffic. The operators can further evaluate the effect of the inter-domain SAVNET on the legitimate traffic and spoofing traffic, such as limiting the rate of all the spoofing traffic without hurting the legitimate traffic.¶

• Third, when the validation accuracy, forwarding performance, and operational overhead have been verified on a large scale by the live network, the inter-domain SAVNET can open the function to directly block the spoofing traffic that is justified by the SAV table in the data plane.¶

10. Convergence Considerations

Convergence issues SHOULD be carefully considered in inter-domain SAV mechanisms due to the dynamic nature of the Internet. Internet routes undergo continuous changes, and SAV rules MUST proactively adapt to these changes, such as prefix and topology changes, in order to prevent false positives and reduce false negatives. To effectively track these changes, the SIM should promptly collect SAV-related information from various SAV information sources and consolidate them in a timely manner.¶

In particular, it is essential for the SAVNET agents to proactively communicate the changes of the SAV-specific Information between ASes and adapt to route changes promptly. However, during the routing convergence process, the traffic paths of the source prefixes can undergo rapid changes within a short period. The changes of the SAV-specific Information may not be communicated in time between ASes to update SAV rules, false positives or false negatives may happen. Such inaccurate validation is caused by the delays in communicating SAV-specific Information between ASes, which occur due to the factors like packet losses, unpredictable network latencies, or message processing latencies. The design of the SAV-specific communication mechanism should consider these issues to reduce the inaccurate validation.¶

Besides, for the inter-domain SAVNET architecture, the potential ways to deal with the inaccurate validation issues during the convergence of the SAV-specific communication mechanism is to consider using the information from RPKI ROA objects and ASPA objects to generate SAV rules until the convergence process of the SAV-specific communication mechanism is finished, since these information is more stable and can help avoid false positives, and thus avoiding the impact to the legitimate traffic.¶

11. Manageability Considerations

It is crucial to consider the operations and management aspects of SAV information sources, the SAV-specific communication mechanism, SIB, SIM, and SAV table in the inter-domain SAVNET architecture. The following guidelines should be followed for their effective management:¶

First, management interoperability should be supported across devices from different vendors or different releases of the same product, based on a unified data model such as YANG [RFC6020]. This is essential because the Internet comprises devices from various vendors and different product releases that coexist simultaneously.¶

Second, scalable operation and management methods such as NETCONF [RFC6241] and syslog protocol [RFC5424] should be supported. This is important as an AS may have hundreds or thousands of border routers that require efficient operation and management.¶

Third, management operations, including default initial configuration, alarm and exception reporting, logging, performance monitoring and reporting for the control plane and data plane, as well as debugging, should be designed and implemented in the protocols or protocol extensions. These operations can be performed either locally or remotely, based on the operational requirements.¶

By adhering to these rules, the management of SAV information sources and related components can be effectively carried out, ensuring interoperability, scalability, and efficient operations and management of the inter-domain SAVNET architecture.¶

12. Security Considerations

In the inter-domain SAVNET architecture, the SAVNET agent plays a crucial role in generating and disseminating SAV-specific messages across different ASes. To safeguard against the potential risks posed by a malicious AS generating incorrect or forged SAV-specific messages, it is important for the SAVNET agents to employ security authentication measures for each received SAV-specific Message. The majour security threats faced by inter-domain SAVNET can be categorized into two aspects: session security and content security. Session security pertains to verifying the identities of both parties involved in a session and ensuring the integrity of the session content. Content security, on the other hand, focuses on verifying the authenticity and reliability of the session content, thereby enabling the identification of forged SAV-specific Messages.¶

The threats to session security include:¶

• Session identity impersonation: This occurs when a malicious router deceitfully poses as a legitimate peer router to establish a session with the targeted router. By impersonating another router, the malicious entity can gain unauthorized access and potentially manipulate or disrupt the communication between the legitimate routers.¶

• Session integrity destruction: In this scenario, a malicious intermediate router situated between two peering routers intentionally tampers with or destroys the content of the relayed SAV-specific Message. By interfering with the integrity of the session content, the attacker can disrupt the reliable transmission of information, potentially leading to miscommunication or inaccurate SAV-related data being propagated.¶

The threats to content security include:¶

• Message alteration: A malicious router has the ability to manipulate or forge any portion of a SAV-specific message. For example, the attacker may employ techniques such as using a spoofed Autonomous System Number (ASN) or modifying the AS Path information within the message. By tampering with the content, the attacker can potentially introduce inaccuracies or deceive the receiving ASes, compromising the integrity and reliability of the SAV-related information.¶

• Message injection: A malicious router injects a seemingly "legitimate" SAV-specific message into the communication stream and directs it to the corresponding next-hop AS. This type of attack can be likened to a replay attack, where the attacker attempts to retransmit previously captured or fabricated messages to manipulate the behavior or decisions of the receiving ASes. The injected message may contain malicious instructions or false information, leading to incorrect SAV rule generation or improper validation.¶

• Path deviation: A malicious router intentionally diverts a SAV-specific Message to an incorrect next-hop AS, contrary to the expected path defined by the AS Path. By deviating from the intended routing path, the attacker can disrupt the proper dissemination of SAV-related information and introduce inconsistencies or conflicts in the validation process. This can undermine the effectiveness and accuracy of source address validation within the inter-domain SAVNET architecture.¶

Overall, inter-domain SAVNET shares similar security threats with BGP and can leverage existing BGP security mechanisms to enhance both session and content security. Session security can be enhanced by employing session authentication mechanisms used in BGP. Similarly, content security can benefit from the deployment of existing BGP security mechanisms like RPKI, BGPsec, and ASPA. While these mechanisms can address content security threats, their widespread deployment is crucial. Until then, it is necessary to develop an independent security mechanism specifically designed for inter-domain SAVNET. One potential approach is for each origin AS to calculate a digital signature for each AS path and include these digital signatures within the SAV-specific messages. Upon receiving a SAV-specific Message, the SAVNET agent can verify the digital signature to ascertain the message's authenticity. Furthermore, it is worth noting that the information channel of the inter-domain SAVNET architecture may need to operate over a network link that is currently under a source address spoofing attack. As a result, it may experience severe packet loss and high latency due to the ongoing attack, and the implementation of the information channel should ensure uninterrupted communication. Detailed security designs and considerations will be addressed in a separate draft, ensuring the robust security of inter-domain SAVNET.¶

13. Privacy Considerations

TBD¶

14. IANA Considerations

This document has no IANA requirements.¶

15. Scope and Assumptions

In this architecture, the choice of protocols used for communication between the SIM and different SAV information sources is not limited. The inter-domain SAVNET architecture presents considerations on how to consolidate SAV-related information from various sources to generate SAV rules and perform SAV using the SAV table in the dataplane. The detailed design and implementation for SAV rule generation and SAV execution depend on the specific inter-domain SAV mechanisms employed.¶

This document does not cover administrative or business agreements that may be established between the involved inter-domain SAVNET parties. These considerations are beyond the scope of this document. However, it is assumed that authentication and authorization mechanisms can be implemented to ensure that only authorized ASes can communicate SAV-related information.¶

This document makes the following assumptions:¶

• All ASes where the inter-domain SAVNET is deployed are assumed to provide the necessary connectivity between SAVNET agent and any intermediate network elements. However, the architecture does not impose any specific limitations on the form or nature of this connectivity.¶

• Congestion and resource exhaustion can occur at various points in the inter-domain networks. Hence, in general, network conditions should be assumed to be hostile. The inter-domain SAVNET architecture must be capable of functioning reliably under all circumstances, including scenarios where the paths for delivering SAV-related information are severely impaired. It is crucial to design the inter-domain SAVNET system with a high level of resilience, particularly under extremely hostile network conditions. The architecture should ensure uninterrupted communication between inter-domain SAVNET agents, even when data-plane traffic saturates the link.¶

• The inter-domain SAVNET architecture does not impose rigid requirements for the SAV information sources that can be used to generate SAV rules. Similarly, it does not dictate strict rules on how to utilize the SAV-related information from diverse sources or perform SAV in the dataplane. Network operators have the flexibility to choose their approaches to generate SAV rules and perform SAV based on their specific requirements and preferences. Operators can either follow the recommendations outlined in the inter-domain SAVNET architecture or manually specify the rules for governing the use of SAV-related information, the generation of SAV rules, and the execution of SAV in the dataplane.¶

• The inter-domain SAVNET architecture does not impose restrictions on the selection of the local AS with which AS to communicate SAV-specific Information. The ASes have the flexibility to establish connections for SAV-specific communication based on the manual configurations set by operators or other automatic mechanisms.¶

• The inter-domain SAVNET architecture provides the flexibility to accommodate Quality-of-Service (QoS) policy agreements between SAVNET-enabled ASes or local QoS prioritization measures, but it does not make assumptions about their presence. These agreements or prioritization efforts are aimed at ensuring the reliable delivery of SAV-specific Information between SAVNET agents. It is important to note that QoS is considered as an operational consideration rather than a functional component of the inter-domain SAVNET architecture.¶

• The SAVNET communication mechanisms are loosely coupled and are used for communicating or gathering SAV-related information, and how the inter-domain SAVNET synchronizes the management and operation configurations is out of scope of this document.¶

16. Contributors

Igor Lubashev
Akamai Technologies
145 Broadway
Cambridge, MA, 02142
United States of America
Email: ilubashe@akamai.com¶

Many thanks to Igor Lubashev for the significantly helpful revision suggestions.¶

Acknowledgements

Many thanks to Alvaro Retana, Kotikalapudi Sriram, Rüdiger Volk, Xueyan Song, Ben Maddison, Jared Mauch, Joel Halpern, Aijun Wang, Jeffrey Haas, Xiangqing Chang, Changwang Lin, Mingxing Liu, Zhen Tan, Yuanyuan Zhang, Yangyang Wang, Antoin Verschuren etc. for their valuable comments on this document.¶

Authors' Addresses