BMP for BGP Route Leak Detection
draft-gu-grow-bmp-route-leak-detection-01

Abstract

According to RFC7908, Route leaks refer to case that the delivery range of route advertisements is beyond the expected range. For many current security protection solutions, the ISPs (Internet Service Providers) are focusing on finding ways to detect the happening of route leaks. However, the real-time route leak detection if any occurs is important as well. This document extends the BGP Monitoring Protocol (BMP) to provide a routing security scheme suitable for ISPs to detect BGP route leaks within their own networks.

Requirements Language

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119.

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 September 12, 2019.

Copyright Notice

Copyright (c) 2019 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 Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

Table of Contents

• 1. Terminology
• 2. Introduction
• 2.1. ISP Route Leak Prevention Methods
• 2.2. Challenge of the Current Route Leak Prevention Methods
• 3. Route Leak Detection Considerations
• 4. Extending BMP for RLD
• 5. Acknowledgements
• 6. IANA Considerations
• 7. Security Considerations
• 8. References
• 8.1. Normative References
• 8.2. Informative References
• Authors' Addresses

1. Terminology

BMP: BGP Monitoring Protocol

BMS: BGP Monitoring Station

C2P: Customer to Provider

ISP: Internet Service Provider

P2P: Peer to Peer

RIB: Routing Information Base

RLD: Route Leak Detection

2. Introduction

RFC 7908 defines "Route Leak" as: A route leak is the propagation of routing announcement(s) beyond their intended scope, which can result in possible situations such as eavesdropping, device overload, route black hole and so on. More specifically, the intended scope of route announcements is usually defined by local route filtering/distribution policies within devices. These policies are designed to realise the pair-wise peering business relationships between ASes (autonomous systems), which include Customer to Provider (C2P), Peer to Peer (Peer to Peer), and Provider to Customer (P2C). In a C2P relationship, the customer pays the provider for traffic sent between the two ASes. In return, the customer gains access to the ASes the provider can reach, including those which the provider reaches through its own providers. In a P2P relationship, the peering ASes gain access to each other's customers, typically without either AS paying the other[Luckie]. RFC 7908 classifies six typical route leaks situations based on the documented events.

2.1. ISP Route Leak Prevention Methods

Since BGP itself does not provide any route leak prevention/protection, in the current networks, network administrators/operators typically configure export policies on the AS border routers (ASBRs) to prevent route leak. For example, refer to the topology in Figure 1, the bussiness relationship between AS2 and AS1 is P2C, and P2C between AS1 and AS3, and C2P between AS1 and AS4. According to RFC 7908, for AS1, any route received from the provider AS (i.e., AS2 here) and then distributed to its provider AS (i.e., AS4) is treated as route leak (Type 1 route leak). Thus, to prevent such case from happening, an export policy is configured at ASBR R2 of AS1. The export strategies are meant for the intention that "routes from AS2 can be sent to AS3, and cannot be sent to AS4." Routes received from AS2 at AS1 (i.e., R1 here) are marked with BGP community attributes so that when these routes arrive at any exit ASBR of AS1 (i.e., R2 here) is filtered by the route leak policy configured at R2 by identifying the community attribute attached from R1. This community attribute stands for the peering business relationship between AS2 and AS1. Suppose the destination of the route A is AS4, then R2 will not distribute Route A to AS4 were the export policies configured correctly.

                  *************************
                  *                       *       "Send Route
   Route A        *          AS1         +*+-----> A to AS3"
   +-->           *                     + *      +-----+
   +-----+        *  +---+         +---+  *+P2C+--|AS3 +----+ ...
+--+ AS2 +---+P2C+*+-+ R1+---------+ R2|  *      +-----+
   +-----+        *  +-+-+\        +---+  *
                  *    |   \\    //  |\   *
                  *    |     \\//    | \  *       "Do not send
                  *    |     //\     |  \+-------> Route A to AS4"
                  *    |   //   \\   |    *      +-----+
                  *    |  /       \  |    *+C2P+--|AS4 +----+ ...
                  *  +-+-+         +-+-+  *      +-----+
                  +--+ R3+---------+ R4|  *
                  *  +---+         +---+  *
                  *                       *
                  *************************

          Figure 1: Route propagatin between ISPs

2.2. Challenge of the Current Route Leak Prevention Methods

However, it could happen that the export policies configured at ASBRs to prevent route leak are misconfigured or simply out of date considering the changes of bussiness relationships between ASes. For example, the export policies at R2 fails to filter Route A and distributes it to AS4, then a route leak happens. Thus, in addition to such route leak prevention methods, there requires a valid detection method to detect any occurred leak in a timely manner so that the incorrect policies can be identified to avoid further leaks.

3. Route Leak Detection Considerations

There are some existing methods proposed for Route Leak Detection (RLD).

It's straightforward to think of the idea of using a route's AS path combined with the business relationship information between ISPs/ASes to detect any route leak. However, there exist implementation difficuties.

First of all, the business relationship information between ISPs/ASes is not publicly disclosed due to confidentiality reasons. Thus, many attempts have been made to infer relationships and strategies between ASs, however, the accuracy of these techniques is often questioned. In particular, the increase in the number of Internet Exchange Points (IXPs) and their role in the recent "flattening" of the Internet topology, makes that a large fraction of AS relationships cannot be discovered using these data collection points [Siddiqui].

Secondly, the acquisition of BGP AS path information is also no easy work. Some BGP monitoring tools, such as Looking Glass and Route View, the data accuracy or completeness remains to be an issue. This has led to the such BGP monitoring tools not being well used by various ISPs.

Some other technologies extend existing routing protocols to realize RLD. For example, modify the BGP update message, which may bring about compatibility problems involved in the implementation of the solution. Besides, new extension brings interoperation, device upgrade issues. Thus, extending the routing protocols is not the first choice for leak detection if there are other options.

Summarizing the above discussions, we have identified the following considerations when designing a RLD solution:

• Consideration 1: The detection should not depend on inferred business relationship information, which leads to inaccurate detection;
• Consideration 2: The detection should not depend on inaccurate/incomplete AS path information , which leads to inaccurate detection or a detection miss;
• Consideration 3: The detection should try to avoid extension works of routing protocols considering the interoperation issues;

BMP (BGP Monitoring Protocol) is currently deployed by OTT and operators to monitor the BGP routes, such as monitoring BGP Adj-RIB-In using the process defined in [RFC7854], and monitoring BGP Adj-RIB-Out using the process defined in [I-D.ietf-grow-bmp-adj-rib-out]. Considering the above mentioned requirements of RLD design, extending BMP to collect the business relationships between an ISP and its neighboring ASes can be a good choice for this single ISP to do RLD. There are several merits:

• First of all, it does not involve data disclosure issue since the collected relationship information is only between itself and its neighboring ASes;
• Secondly, BMP provides accurate and complete BGP data monitoring within a singe AS;
• Thirdly, it does not require BGP extension work, and thus no interoperation concern.

Thus, a single ISP can deploy this method to do RLD without relying on any other information from either other ISPs or third party tools.

4. Extending BMP for RLD

                            +------------+
                            | BMP server |
                     +------>      +     +<-------+
                     |      | RLD ser^er |        |
                     +      +------------+        +
             BMP RM adj_rib_in:              BMP RM adj_rib_out:
             relationship between            relationship between
             AS2 and AS1                     AS1 and AS4
                     |                            +
                  ***|*********************       |
                  *  |                    *       |   "Send Route
   Route A        *  |       A 1         +*+---------> A to AS3"
   +-->           *  |                  + *       |  +-----+
   +-----+        *  +---+         +---+  *+P2C+-----+ AS3 +----+ ...
+--+ AS2 +---+P2C+*+-+ R1+---------+ R2|  *       |  +-----+
   +-----+        *  +-+-+\        +---+  *       |
                  *    |   \\    //  |\   *       |
                  *    |     \\//    | \  *       |   "Do not send
                  *    |     //\     |  \+-----------> Route A to AS4"
                  *    |   //   \\   |    *       |  +-----+
                  *    |  /       \  |    *+C2P+-----+ AS4 +----+ ...
                  *  +-+-+         +-+-+  *       |  +-----+
                  +--+ R3+---------+ R4+----------+
                  *  +---+         +---+  *
                  *                       *
                  *************************

            Figure 2: RLD depolyment by a single ISP

A Relationship TLV is defined for BMP Route Monitoring Message. Considering that the AS relationships are sometims per route based instead of per peer/AS based, this TLV is added at the end of each BGP Update Message, and then wrapped up by the BMP per peer header and comon header. The TLV format is defined as follows:

       +-------------------------------+
       |      Type     |    Value      |
       +-------------------------------+

         Figure 3: Relationship TLV

Type = TBD, the Relatiship TLV indicates that this TLV represents the business relationship between the AS that sends the route and the AS that receives the route.

The Value field is a 2 bit field, and can be "00", "01", and "10", which represents three types of relationships, i.e., P2C, P2P, C2P, respectively.

As shown in Figure 2, with the Relationship TLV attached to each Route Monitoring Message, the RLD server (also working as the BMP server) combines the BMP adj_rib_in message collected from R1 and the BMP adj_rib_out message collected from R4 to decide if there's a route leak. For example, if the Relationship TLV in R1's adj_rib_in message indicates a value of "00", and the Relationship TLV in R4's adj_rib_out message indicates a value of "10", then the RLD server knows there exists a route leak.

5. Acknowledgements

TBD.

6. IANA Considerations

TBD.

7. Security Considerations

TBD.

8. References

8.1. Normative References

8.2. Informative References

Authors' Addresses