Framework of Forwarding Commitment BGP

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 6 January 2024.¶

Copyright Notice

Copyright (c) 2023 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

The fundamental cause of the path manipulation attacks in Internet inter-domain routing is that the de facto Border Gateway Protocol (BGP) [RFC4271] does not have built-in mechanisms to authenticate routing announcements. As a result, an adversary can announce virtually arbitrary paths to a prefix while the network cannot effectively verify the authenticity of the route announcements.¶

In addition to the lack of control plane authentication, ensuring that the actual forwarding paths in the dataplane comply with the control plane decisions is also missing in today's inter-domain routing system. This fundamentally limits ASes from filtering unwanted traffic which takes an unauthorized AS path.¶

The representative schemes to secure inter-domain routing are RPKI [RFC6480] and BGPsec [RFC8205]. RPKI provides a foundation for validating the origins of BGP routes. Meanwhile, BGPsec directly builds the path authentication of BGP routes into the BGP path construction itself, where an AS is required to iteratively verify the signatures of each prior AS hop before extending the verification chain with its own approval. As a result, a single legacy or malicious AS can terminate the verification chain, preventing the downstream ASes from reinstating the verification process. This creates the well-known chicken-and-egg problem where the early adopters receive no deployment benefits which further limits new adoption.¶

Finally, due to the lack of practical protocols to check the consistency between the dataplane forwarding and control-plane decisions, enforcing path authorization in the inter-domain forwarding has been not possible to date.¶

This document specifies a framework named FC-BGP, an incrementally deployable security augment to the Internet inter-domain routing and forwarding. FC-BGP relies on the Resource Public Key Infrastructure (RPKI) certificates that attest to the allocation of AS number and IP address resources. To support FC-BGP, a BGP speaker needs to possess a private key associated with an RPKI router certificate [RFC8209] that corresponds to the BGP speaker's AS number.¶

2. Threat Model and Assumptions

We assume that ASes participating in FC-BGP have access to RPKI, which stores authoritative information about the mapping between AS numbers and their owned IP prefixes, as well as ASes' public keys. Given the above assumptions, we consider the following adversary:¶

1. On the control plane, the adversary can launch path manipulation attacks. This means that the adversary will try to manipulate the routing paths to victim ASes or prefixes by sending bogus BGP updates. For example, the adversary might try to reroute traffic to the victim ASes/prefixes through ASes that they control, in order to perform (encrypted) traffic analysis.¶
2. On the dataplane, the adversary can spoof source addresses and send unwanted network traffic to the victim ASes prefixes.¶

3. Overview

          |     1.BGP Announcement Path        |
          +---------------------+------------->|
Control   |            ^        |              |
Plane     |    1.1     |        |    1.2       |
          | Generation |        v Validation   |
     +----+---+      +-+----------+       +----+---+
     |        |      | Forwarding |       |        |
     |  AS A  |      | Commitment |       |  AS B  |
     |        |      |    (FC)    |       |        |
     +----+---+      +----------+-+       +----+---+
          |            ^        |              |
 Data     |    2.2     |        |    2.1       |
 Plane    | Validation |        |  Binding     |
          |            |        v              |
          |<-----------+-----------------------+
          |         2.Forwarding Path          |

An overview of FC-BGP is shown in Figure 1. The key primitive in FC-BGP is the Forwarding Commitment (FC), which is a publicly verifiable code that certifies an AS's routing intent on one of its directly connected hops, i.e., an FC indicates whether the AS is willing to carry traffic for a specific prefix over the hop. Upon receiving a BGP announcement, if AS A decides to accept the route and extends the path to its (selected) neighbor AS B, AS A commits its routing intent by generating a cryptographically-signed FC. Therefore, downstream on-path ASes (such as AS B) can validate the correctness of a BGP update by checking the FCs associated with the individual hops on the AS-path. Because the FCs are designed to be hop-specific and path-agnostic, a deployed AS can immediately certify its routing intent regardless of the deployment status of other ASes. This is fundamentally different from existing path-level BGP authentication protocol (e.g., BGPsec) where an on-path AS cannot approve any form of routing intent unless all on-path ASes are upgraded.¶

FC-BGP is not bound to a specific FC propagation method and can use, but is not limited to, the following mechanisms:¶

1. Extend BGP Update Message. Assign a new BGP Update Path Attribute to carry FCs.¶
2. Propose a new propagation protocol that guarantees consistent FC propagation.¶
3. Use existing network infrastructure, such as extending RPKI to add a new signed object to store FCs.¶

Meanwhile, the flexibility of FCs further enables efficient forwarding validation on the dataplane. Specifically, because the FCs are self-proving, an AS can conceptually construct a certified AS-path using a list of consecutive per-hop FCs, and then binds its network traffic (identified by < src-AS, dst-AS, prefix >) to the path, such as < AS B, AS A, P(B) >, where P(B) is the prefix owned by AS B destined to AS A. This binding information essentially defines the authorized forwarding path for the traffic < AS B, AS A, P(B) >. Therefore, by advertising the binding information globally, both on-path and off-path ASes are aware of the desired forwarding paths so that they can collaboratively discard unwanted traffic that takes unauthorized paths.¶

Similar to FC propagation, the propagation of binding messages in FC-BGP is not restricted to specific methods and can be, but is not limited to, the following:¶

1. Propose a new propagation protocol that guarantees the consistency of binding messages.¶
2. Use existing network infrastructure, such as extending RPKI to add a new signed object to store binding messages.¶

4. Forwarding Commitment

FC-BGP enhances the security of inter-domain routing and forwarding by building a publicly verifiable view of the forwarding commitments. At a high level, a routing commitment (FC) of an AS is a cryptographically-signed primitive that binds the AS's routing decisions (e.g. willing to forward traffic for a prefix via one of its directly-connected hops). With this view, ASes are able to:¶

1. Evaluate the authenticity (or security) of the control plane BGP announcements based on committed routing decisions from relevant ASes.¶
2. Ensure that the dataplane forwarding is consistent with the routing decisions committed in the control plane.¶

Upon receiving a BGP announcement, an upgraded AS generates a corresponding FC that contains the core information of the announcement, such as prefixes, sending AS, and receiving AS, and will be signed with the sender's private key for security.¶

5. BGP Path Validation

                                            FClist:F(C,D,P)
                          FClist:F(B,C,P)          F(B,C,P)
       FClist:F(A,B,P)           F(A,B,P)          F(A,B,P)
     | AS Path:A       |  AS Path:A-B     | AS Path:A-B-C    |
     +---------------->+----------------->+----------------->|
     |                 |                  |                  |
     |                 |                  |                  |
     |                 |                  |                  |
+----+---+        +----+---+         +----+---+         +----+---+
|  AS A  +--------+  AS B  +---------+  AS C  +---------+  AS D  |
+----+---+        +----+---+         +----+---+         +----+---+
     |                 |                  |                  |
     |        F(C,D,P)-(src:P(D),dst:P(A))+<-----------------+
     |                 |                  |                  |
     |                 |                  |                  |
     |                 |<-----------------+------------------+
     |                 |      F(B,C,P)-(src:P(D),dst:P(A))   |
     |                 |                  |                  |
     |                 |                  |                  |
     |<----------------+------------------+------------------+
                   F(A,B,P)-(src:P(D),dst:P(A))

Consider an illustrative example using the four-AS topology shown in Figure 2. In this process, FC-BGP generates the corresponding FC and propagates to downstream ASes (e.g., adding it to the Path Attributes of the BGP updates), so that the receiving AS can validate the authenticity of the announcement. Suppose AS C receives a BGP announcement P(A): A->B->C from its neighbor B. If AS C decides to further advertise this path to its neighbor D based on its routing policy, it generates a FC F(C,D,P), propagates it to AS D, and forwards the BGP update message to D.¶

When AS D receives the route from C, it can determine the authenticity of the current AS path by verifying the list of FCs correctly reflects the AS path.¶

6. Forwarding Validation

AS shown in Figure 2, to enable forwarding validation, ASes need to announce the traffic-FCs binding relationships. Specifically, suppose AS D confirms that the AS-path C->B->A for reaching prefix P(A) is legitimate, it binds the traffic (src:P(D),dst:P(A)) (where P(D) is the prefix owned by AS D) with the FC list F(C,D,P)-F(B,C,P)-F(A,B,P), and then publicly announces the binding relationship.¶

Upon receiving the relationship, other ASes can build traffic filtering rules based on the relationship to enable forwarding validation on dataplane. For instance, by interpreting the binding relationship produced by AS D, AS C confirms that the traffic (src:P(D), dst:P(A)) shall be forwarded over the link L(CD), and AS B confirms that the traffic shall be forwarded on link L(BC). Network traffic violating these binding rules is considered to take unauthorized paths.¶

To enable network-wide forwarding verification, these binding rules may be further broadcast globally (instead of just informing the ASes on the AS-path) so that off-path ASes can also discard unauthorized flows.¶

7. Security Considerations

8. IANA Considerations

This document has no IANA actions.¶

Acknowledgments

TODO acknowledge.¶

Authors' Addresses