Internet Engineering Task Force (IETF) V. Manral Request for Comments: 6039 IP Infusion Category: Informational M. Bhatia ISSN: 2070-1721 Alcatel-Lucent J. Jaeggli Nokia Inc. R. White Cisco Systems October 2010 Issues with Existing Cryptographic Protection Methods for Routing Protocols Abstract Routing protocols have been extended over time to use cryptographic mechanisms to ensure that data received from a neighboring router has not been modified in transit and actually originated from an authorized neighboring router. The cryptographic mechanisms defined to date and described in this document rely on a digest produced with a hash algorithm applied to the payload encapsulated in the routing protocol packet. This document outlines some of the limitations of the current mechanism, problems with manual keying of these cryptographic algorithms, and possible vectors for the exploitation of these limitations. Status of This Memo This document is not an Internet Standards Track specification; it is published for informational purposes. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Not all documents approved by the IESG are a candidate for any level of Internet Standard; see Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6039.
Copyright Notice Copyright (c) 2010 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 (http://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. Problem Statement ...............................................3 1.1. Pre-Image vs. Collision Attacks ............................4 1.2. Concerns about MD5 and the SHA-1 Algorithm .................4 2. Open Shortest Path First Version 2 (OSPFv2) .....................5 2.1. Management Issues with OSPFv2 ..............................5 2.2. Technical Issues with OSPFv2 ...............................6 3. Open Shortest Path First Version 3 (OSPFv3) .....................7 3.1. Management Issues with OSPFv3 ..............................7 3.2. Technical Issues with OSPFv3 ...............................8 4. Intermediate System to Intermediate System Routing Protocol (IS-IS) ................................................9 4.1. Management Issues with IS-IS ...............................9 4.2. Technical Issues with IS-IS ...............................10 5. Border Gateway Protocol (BGP-4) ................................11 5.1. Management Issues with BGP-4 ..............................12 5.2. Technical Issues with BGP-4 ...............................13 6. The Routing Information Protocol (RIP) .........................13 6.1. Technical Issues with RIP .................................14 7. Bidirectional Forwarding Detection (BFD) .......................15 7.1. Technical Issues with BFD .................................15 8. Security Considerations ........................................17 9. Acknowledgements ...............................................17 10. References ....................................................17 10.1. Normative References .....................................17 10.2. Informative References ...................................18 11. Contributor's Address .........................................21
1. Problem Statement Protocols, such as OSPF version 2 [RFC2328], version 3 [RFC5340], IS-IS [RFC1195], BGP-4 [RFC4271], and BFD [RFC5880], employ various mechanisms to create a cryptographic digest of each transmitted protocol packet. Traditionally, these digests are the results of a one-way hash algorithm, such as Message Digest 5 (MD5) [RFC1321], across the contents of the packet being transmitted. A secret key is used as the hash base (or seed). The digest is then recomputed by the receiving router, using the same key as the original router used to create the hash, then compared with the transmitted digest to verify: o That the router originating this packet is authorized via the shared key mechanism to peer with the local router and exchange routing data. The implicit trust of the routing protocol exchange protected by a shared secret is intended to protect against the injection of falsely generated routing data into the routing system by unauthorized systems. o That the data has not been altered in transit between the two neighboring routers. Digest verification schemes are not intended to protect the confidentiality of information being exchanged between routers. The information (entries in the routing table) is potentially available through other mechanisms. Moreover, access to the physical media between two routers exchanging routing data will confer the ability to capture or otherwise discover the contents of the routing tables in those routers. Authentication mechanisms defined today have notable limitations: o Manual configuration of shared secret keys, especially in large networks and between networks, poses a major management problem. In many cases, it is challenging to replace keys without significant coordination or disruption. o In some cases, when manual keys are configured, some forms of replay protection are no longer possible, allowing the routing protocol to be attacked through the replay of captured routing messages. This document outlines some of the problems with manual keying of these cryptographic algorithms.
1.1. Pre-Image vs. Collision Attacks A pre-image attack (an attempt to find new data with the same hash value) would enable someone to find an input message that causes a hash function to produce a particular output. In contrast, a collision attack finds two messages with the same hash, but the attacker can't pick what the message will be. Feasible collision attacks against MD4, MD5, HAVAL-128, and RIPEMD have been documented in [Crypto2004]. The ability to produce a collision does not currently introduce any obvious or known attacks on routing protocols. Pre-image attacks have the potential to cause problems in the future; however, due to the message length, there are serious limitations to the feasibility of mounting such an attack. Protocols themselves have some built-in protection against collision attacks. This is because a lot of values for fields in a protocol packet are invalid or will produce an unusable packet. For example, in OSPF the Link State Advertisement (LSA) type can be from 1 to 11. Any other value in the field will result in the packet being discarded. Assume two packets M and M' are generated and have the same hash. The above condition will further reduce the ability to produce a message that is also a correct message from the protocol perspective, as a lot of potential values are themselves not valid. 1.2. Concerns about MD5 and the SHA-1 Algorithm There are published concerns about the overall strength of the MD5 algorithm ([Dobb96a], [Dobb96b], [Wang04]). While those published concerns apply to the use of MD5 in other modes (e.g., use of MD5 X.509v3/PKIX digital certificates), they are not an attack upon Keyed MD5 and Hash-based Message Authentication Code MD5 (HMAC-MD5), which is what the current routing protocols have specified. There are also published concerns about the Secure Hash Algorithm (SHA) algorithm ([Wang05], [Philip01], [Prav01], [Prav02], [Arjen05]) and also concerns about the MD5 and SHA algorithms in the HMAC [RFC2104] mode ([RR07], [RR08]). The National Institute of Standards and Technology (NIST) will be supporting HMAC-SHA-1 even after 2010 [NISTHmacSHA], whereas it will drop support for SHA-1 in digital signatures. NIST also recommends application and protocol designers move to the SHA-2 family of hash functions (i.e., SHA-224, SHA-256, SHA-384 and SHA-512) for all new applications and protocols.
However, as explained above, such attacks are currently not applicable to the routing protocols. Separately, some organizations (e.g., the US government) prefer NIST algorithms, such as the SHA family, over other algorithms (like MD5) for local policy reasons. 2. Open Shortest Path First Version 2 (OSPFv2) OSPF [RFC2328] describes the use of an MD5 digest with OSPF packets. MD5 keys are manually configured. The OSPF packet header includes an authentication type field as well as 64 bits of data for use by the appropriate authentication scheme. OSPF also provides for a non- decreasing sequence number to be included in each OSPF protocol packet to protect against replay attacks. "OSPF with Digital Signatures" [RFC2154] is an Experimental RFC that describes extensions to OSPF to digitally sign its Link State Advertisements (LSAs). It is believed that if stronger authentication and security is required, then OSPF (and the other routing protocols) must migrate to using full digital signatures. Doing this would enable precise authentication of the OSPF router originating each OSPF link-state advertisement, and thereby provide much stronger integrity protection for the OSPF routing domain. However, since there have been no deployments, there is precious little operational experience with this specification, and hence it is not covered in this document. 2.1. Management Issues with OSPFv2 According to the OSPF specification [RFC2328], digests are applied to packets transmitted between adjacent neighbors, rather than being applied to the routing information originated by a router (digests are not applied at the LSA level, but rather at the packet level). [RFC2328] states that any set of OSPF routers adjacent across a single link may use a different key to build MD5 digests than the key used to build MD5 digests on any other link. Thus, MD5 keys may be configured, and changed, on a per-link basis in an OSPF network. OSPF does not specify a mechanism to negotiate keys, nor does it specify any mechanism to negotiate the hash algorithms to be used. With the proliferation of the number of hash algorithms, as well as the need to continuously upgrade the algorithms, manually configuring the information becomes very tedious. It should also be noted that rekeying OSPF requires coordination among the adjacent routers.
2.2. Technical Issues with OSPFv2 While OSPF provides relatively strong protection through the inclusion of MD5 digests, with additional data and sequence numbers in transmitted packets, there are still attacks against OSPF: o The sequence number is initialized to zero when forming an adjacency with a newly discovered neighbor. When an adjacency is brought down, the sequence number is also set to zero. If the cryptographically protected packets of a router that is brought down (for administrative or other reasons) are replayed by a malicious router, traffic could be forced through the malicious router. A malicious router might then induce routing loops, or intercept or blackhole the traffic. o OSPF allows multiple packets with the same sequence number. This means that it's possible to replay the same packet many times before the next legitimate packet is sent. An attacker may resend the same packet repeatedly until the next Hello packet is transmitted and received. The Hello interval, which is unknown, determines the attack window. o OSPF does not require the use of any particular hash algorithm; however, only the use of MD5 digests for authentication and replay protection is specified in RFC 2328. Most OSPF implementations only support MD5 in addition to Null and Simple Password authentication. Recently, limitations in collision-resistance properties of the MD5 and SHA-1 hash functions have been discovered; [RFC4270] summarizes the discoveries. There have been attacks against the use of MD5 as a hash; while these attacks do not directly apply to the use of MD5 in routing protocols, it is prudent to have other options available. For this reason, the general use of these algorithms should be discouraged, and [RFC5709] adds support for using SHA-1 and SHA-2 with the HMAC construct for OSPF. o OSPF on a broadcast network shares the same key between all neighbors on that broadcast network. Some OSPF packets are sent to a multicast address. Spoofing by any malicious neighbor possessing credentials or replayable packets is therefore very easy. Possession of the key itself is used as an identity validation, and no other identity check is used. A malicious neighbor could send a packet, forging the identity of the sender as being from another neighbor. There would be no way in which the victim could distinguish the identity of the packet sender.
o In some OSPF implementations, neighbors on broadcast, non- broadcast multi-access (NBMA), and point-to-multipoint networks are identified by the IP address in the IP header. The IP header is not covered by the MAC in the cryptographic authentication scheme as described in RFC 2328, and an attack can be made to exploit this omission. Assume the following scenario. R1 sends an authenticated HELLO to R2. This HELLO is captured and replayed back to R1. The source IP in the IP header of the replayed packet is changed to that of R2. R1, not finding itself in the HELLO, would deduce that the connection is not bidirectional and would bring down the adjacency. 3. Open Shortest Path First Version 3 (OSPFv3) OSPFv3 [RFC5340] relies on the IP Authentication Header (AH) [RFC4302] and the IP Encapsulating Security Payload (ESP) [RFC4303] to cryptographically sign routing information passed between routers. When using ESP, the null encryption algorithm [RFC2410] is used, so the data carried in the OSPFv3 packets is signed, but not encrypted. This provides data origin authentication for adjacent routers, and data integrity (which gives the assurance that the data transmitted by a router has not changed in transit). However, it does not provide confidentiality of the information transmitted; this is acceptable because the privacy of the information being carried in the routing protocols need not be kept secret. "Authentication/Confidentiality for OSPFv3" [RFC4552] mandates the use of ESP with null encryption for authentication and also does encourage the use of confidentiality to protect the privacy of the routing information transmitted, using ESP encryption. However, it only specifies the use of manual keying of routing information as discussed in the following section. 3.1. Management Issues with OSPFv3 The OSPFv3 security document ("Authentication/Confidentiality for OSPFv3" [RFC4552]) discusses, at length, the reasoning behind using manually configured keys, rather than some automated key management protocol such as IKEv2 [RFC4306]. The primary problem is the lack of a suitable key management mechanism, as OSPF adjacencies are formed on a one-to-many basis and most key management mechanisms are designed for a one-to-one communication model. This forces the
system administrator to use manually configured security associations (SAs) and cryptographic keys to provide the authentication and, if desired, confidentiality services. Regarding replay protection, [RFC4552] states that: Since it is not possible using the current standards to provide complete replay protection while using manual keying, the proposed solution will not provide protection against replay attacks. In the OSPFv3 case, the primary administrative issue with manually configured SAs and keys is the management issue -- maintaining shared sets of keys on all routers within a network. As with OSPFv2, rekeying is an infrequent event requiring coordination. [RFC4552] does not require that all OSPFv3 routers have the same key configured for every neighbor, so each set of neighbors connected to a given link could have a different key configured. While this makes it easier to change the keys (by forcing the system administrator to only change the keys on the routers on a single link), the process of manual configuration for all the routers in a network to change the keys used for OSPFv3 digests and confidentiality on a periodic basis can be difficult. 3.2. Technical Issues with OSPFv3 The primary technical concern with the current specifications for OSPFv3 is that when manual SA and key management is used as specified in "Sequence Number Generation", Section 3.3.2 of [RFC4302]: "The sender assumes anti-replay is enabled as a default, unless otherwise notified by the receiver (see Section 3.4.3) or if the SA was configured using manual key management". Replaying OSPFv3 packets can induce several failures in a network, including: o Replaying Hello packets with an empty neighbor list can cause all the neighbor adjacencies with the sending router to be reset, disrupting network communications. o Replaying Hello packets from early in the designated router election process on broadcast links can cause all the neighbor adjacencies with the sending router to be reset, disrupting network communications. o Replaying database description (DB-Description) packets can cause all FULL neighbor adjacencies with the sending router to be reset, disrupting network communications.
o Replaying link state request (LS-Request) packets can cause all FULL neighbor adjacencies with the sending router to be reset, disrupting network communications. o Capturing a full adjacency process (from two-way all the way to FULL state), and then replaying this process when the router is no longer attached can cause a false adjacency to be formed, allowing an attacker to attract traffic. o OSPFv3 on a broadcast network shares the same key between all neighbors on that network. Some OSPF packets are sent to a multicast address. Spoofing by a malicious neighbor is very easy. Possession of the key itself is used as an identity check. There is no other identity check used. A neighbor could send a packet specifying the packet came from some other neighbor and there would be no way in which the attacked router could figure out the identity of the packet sender. 4. Intermediate System to Intermediate System Routing Protocol (IS-IS) Integrated IS-IS [RFC1195] uses HMAC-MD5 authentication with manual keying, as described in [RFC5304], and has recently been extended to provide support for using the HMAC construct along with the SHA family of cryptographic hash functions [RFC5310]. There is no provision within IS-IS to encrypt the body of a routing protocol message. 4.1. Management Issues with IS-IS [RFC5304] states that each Link State Protocol Data Unit (LSP) generated by an intermediate system is signed with the HMAC-MD5 algorithm using a key manually defined by the network administrator. Since authentication is performed on the LSPs transmitted by an intermediate system, rather than on the packets transmitted to a specific neighbor, it is implied that all the intermediate systems within a single flooding domain must be configured with the same key in order for authentication to work correctly. The initial configuration of manual keys for authentication within an IS-IS network is simplified by a state where LSPs containing HMAC-MD5/HMAC-SHA authentication TLVs are accepted by intermediate systems without the keys, but the digest is not validated. Once keys are configured on all routers, changing those keys becomes much more difficult.
IS-IS [RFC1195] does not specify a mechanism to negotiate keys, nor does it specify any mechanism to negotiate the hash algorithms to be used. With the proliferation of available hash algorithms, as well as the need to upgrade the algorithms, manual configuration requires coordination among intermediate systems, which can become tedious. 4.2. Technical Issues with IS-IS [RFC5304] states: "This mechanism does not prevent replay attacks; however, in most cases, such attacks would trigger existing mechanisms in the IS-IS protocol that would effectively reject old information". The few cases where existing mechanisms in the IS-IS protocol would not effectively reject old information are: - the Hello packets or the IS-IS Hellos (IIHs) that are used to discover neighbors, and - the Sequence Number Packets (SNPs). As described in IS-IS [RFC1195], a list of known neighbors is included in each Hello transmitted by an intermediate system to ensure two-way communications with any specific neighbor before exchanging link state databases. IS-IS does not provide a sequence number. IS-IS packets are vulnerable to replay attacks; any packet can be replayed at any point of time. So long as the keys used are the same, protocol elements that would not be rejected will affect existing sessions. A Hello packet containing a digest within a TLV and an empty neighbor list could be replayed, resulting in all adjacencies with the original transmitting intermediate system to be restarted. A replay of an old Complete Sequence Number Packet (CSNP) could cause LSPs to be flooded, resulting in an LSP storm. IS-IS specifies the use of the HMAC-MD5 and HMAC-SHA-1 to protect IS-IS packets. IS-IS does not have a notion of Key ID. During key rollover, each message received has to be checked for integrity against all keys that are valid. A denial-of-service (DoS) attack may be induced by sending IS-IS packets with random hashes. This will cause the IS-IS packet to be checked for authentication with all possible keys,
increasing the amount of processing required. This issue, however, has been fixed in the recent [RFC5310], which introduces the concept of Key IDs in IS-IS. Recently, limitations in collision-resistance properties of the MD5 and SHA-1 hash functions have been discovered; [RFC4270] summarizes the discoveries. There have been attacks against the use of MD5 as a hash; while these attacks do not directly apply to the use of HMAC-MD5 in IS-IS, it is prudent to have other options available. For this reason, the general use of these algorithms should be discouraged, and [RFC5310] adds support for using HMAC-SHA with IS-IS. IS-IS on a broadcast network shares the same key between all neighbors on that network. This makes spoofing by a malicious neighbor easy since IS-IS packets are sent to a link-layer multicast address. Possession of the key itself is used as an authorization check. A neighbor could send a packet spoofing the identity of a neighbor, and there would be no way in which the attacked router could discern the identity of the malicious packet sender. The Remaining Lifetime field in the LSP is not covered by the authentication. An IS-IS router can receive its own self-generated LSP segment with zero lifetime remaining. In that case, if it has a copy with non-zero lifetime, it purges that LSP, i.e., it increments the current sequence number and floods all the segments again. This is much worse in IS-IS than in OSPF because there is only one LSP other than the pseudonode LSPs for the LANs on which the IS-IS router is the Designated Intermediate System (DIS). In this way, an attacker can force the router to flood all segments -- potentially a large number if the number of routes is large. It also causes the sequence number of all the LSPs to increase fast. If the sequence number increases to the maximum (0xFFFFFFFF), the IS-IS process must shut down for around 20 minutes (the product of MaxAge + ZeroAgeLifetime) to allow the old LSPs to age out of all the router databases. 5. Border Gateway Protocol (BGP-4) BGP-4 [RFC4271] uses TCP [RFC0793] for transporting routing information between BGP speakers that have formed an adjacency. [RFC2385] describes the use of the TCP MD5 digest option for providing packet origin authentication and data integrity protection of BGP packets. [RFC3562] provides suggestions for choosing the key
length of the ad hoc Keyed MD5 mechanism specified in [RFC2385]. There is no provision for confidentiality for any of these BGP messages. TCP MD5 [RFC2385] has recently been obsoleted by a new TCP Authentication Option (TCP-AO) [RFC5925]. [RFC5925] specifies the use of stronger Message Authentication Codes (MACs), protects against replays even for long-lived TCP connections, and provides more details than TCP-MD5 on the association of security with TCP connections. It allows rekeying during a TCP connection, assuming that an out-of-band protocol or manual mechanism provides the new keys. Note that TCP MD5 does not preclude rekeying during a connection, but does not require its support either. Further, TCP-AO supports key changes with zero segment loss, whereas key changes in TCP MD5 can lose segments in transit during the changeover or require trying multiple keys on each received segment during key use overlap because TCP MD5 lacks an explicit Key ID. Although TCP recovers lost segments through retransmission, loss can have a substantial impact on performance. However, this document covers only TCP MD5, as all current deployments are still using BGP with TCP MD5 and have not upgraded to [RFC5925]. There isn't enough operational experience present to evaluate the technical and management issues with this proposal yet. Compared to previously described IGP protocols, BGP has additional exposure due to the nature of the environment where it is typically used -- namely, between autonomous networks (under different administrative control). While routers running interior gateway protocols may all be configured with the same administrative authority, two BGP peers may be in different administrative domains, having different policies for key strength, rollover frequency, etc. An autonomous system must often support a large number of keys at different BGP boundaries, as each connecting AS represents a different administrative entity. In practice, once set, shared secrets between BGP peers are rarely, if ever, changed. 5.1. Management Issues with BGP-4 Each pair of BGP speakers forming a peering may have a different MD5 shared key that facilitates the independent configuration and changing of keys across a large-scale network. Manual configuration and maintenance of cryptographic keys across all BGP sessions is a challenge in any large-scale environment. Most BGP implementations will accept BGP packets with a bad digest up to the hold interval negotiated between BGP peers at peering startup, in order to allow for MD5 keys to be changed with minimal impact on
operation of the network. This technique does, however, allow some short period of time during which an attacker may inject BGP packets with false MD5 digests into the network and can expect those packets to be accepted, even though the MD5 digests are not valid. 5.2. Technical Issues with BGP-4 BGP relies on TCP [RFC0793] for transporting data between BGP speakers. BGP can rely on TCP's protection against data corruption and replay to preclude replay attacks against BGP sessions. A great deal of research has gone into the feasibility of an attacker overcoming these protections, including [TcpWindow] and [Conv01]. Most router and operating system (OS) vendors have modified their TCP implementations to resolve the security vulnerabilities described in these references, where possible. As mentioned earlier, MD5 is vulnerable to collision attacks and can be attacked through several means, such as those explored in [Wang04]. Though it can be argued that the collision attacks do not have a practical application in this scenario, the use of MD5 should be discouraged. Routers performing cryptographic processing of packets in software may be exposed to additional opportunities for DoS attacks. An attacker may be able to transmit enough spoofed traffic with false digests that the router's processor and memory resources are consumed, causing the router to be unable to perform normal processing. This exposure is particularly problematic between routers not under unified administrative control. 6. The Routing Information Protocol (RIP) The initial version of RIP was specified in STD 34 [RFC1058]. This version did not provide for any authentication or authorization of routing data, and thus was vulnerable to any of a number of attacks against routing protocols. This limitation was one reason why this protocol was moved to Historic status [RFC1923]. RIPv2, originally specified in [RFC1388], then [RFC1723], was finalized in STD 56 [RFC2453]. This version of the protocol provides for authenticating packets with a digest. The details thereof have initially been provided in "RIP-2 MD5 Authentication" [RFC2082]; "RIPv2 Cryptographic Authentication" [RFC4822] obsoletes [RFC2082] and adds details of how the SHA family of hash algorithms can be used to protect RIPv2. [RFC2082] only specified the use of Keyed MD5.
6.1. Technical Issues with RIP o The sequence number used by a router is initialized to zero at startup, and is also set to zero whenever the neighbor is brought down. If the cryptographically protected packets of a router that is brought down (for administrative or other reasons) are stored by a malicious router, the new router could replay the packets from the previous session, thus forcing traffic through the malicious router. Dropping of such packets by the router could result in blackholes. Also, forwarding wrong packets could result in routing loops. o RIPv2 allows multiple packets with the same sequence number. This could mean the same packet may be replayed many times before the next legitimate packet is sent. An attacker may resend the same packet repeatedly until the next Hello packet is transmitted and received, which means the Hello interval therefore determines the attack window. o RIPv2 [RFC2453] did not specify the use of any particular hash algorithm. RFC 4822 introduced HMAC-SHA1 as mandatory to implement, along with Keyed MD5 as specified in [RFC2082]. Support for Keyed MD5 was mandated to ensure compatability with legacy implementations. o "RIPv2 Cryptographic Authentication" [RFC4822] does not cover the UDP and the IP headers. It is therefore possible for an attacker to modify some fields in the above headers without routers becoming aware of it. There is limited exposure to modification of the UDP header, as the RIP protocol uses only it to compute the length of the RIP packet. Changes introduced in the UDP header would cause RIP authentication to fail the RIP authentication, thereby limiting exposure. RIP uses the source IP address from the IP header to determine which RIP neighbor it has learnt the RIP Update from. Changing the source IP address can be used by an attacker to disrupt the RIP routing sessions between two routers R1 and R2, as shown in the following examples. Scenario 1: R1 sends an authenticated RIP message to R2 with a cryptographic sequence number X.
The attacker then needs a packet with a higher sequence number originated by R2 either, from this session or from some earlier session. The attacker can then replay this packet to R2 by changing the source IP to that of R1. R2 would then no longer accept any more RIP Updates from R1, as those would have a lower cryptographic sequence number. After 180 seconds (or less), R2 would consider R1 timed out and bring down the RIP session. Scenario 2: R1 announces a route with cost C1 to R2. This packet can be captured by an attacker. Later, if this cost changes and R1 announces this with a different cost C2, the attacker can replay the captured packet, modifying the source IP to a new arbitrary IP address, thereby masquerading as a different router. R2 will accept this route and the router as a new gateway, and R2 would then use the non-existent router as a next hop for that network. This would only be effective if the cost C1 is less than C2. 7. Bidirectional Forwarding Detection (BFD) BFD is specified in [RFC5880]. Extensions to BFD for multihop [RFC5883] and single hop [RFC5881] are defined for IPv4 and IPv6. It is designed to detect failure with the forwarding plane next hop. The BFD base specification specifies an optional authentication mechanism that can be used by the receiver of a packet to be able to authenticate the source of the packet. It relies on the facts that the keys are shared between the peers and no mechanism is defined for the actual key generation. 7.1. Technical Issues with BFD o The level of security provided is based on the Authentication Type used. However, the authentication algorithms defined are MD5 or SHA-1 based. As mentioned earlier, MD5 and SHA-1 are both known to be vulnerable to collision attacks. o The BFD specification mentions mechanisms to allow for the change of authentication state based on the state of a received packet. This can cause a denial-of-service attack where a malicious authenticated packet (stored from a past session) can be relayed
over a session that does not use authentication. This causes one end to assume that authentication is enabled at the other end, and hence the BFD adjacency is dropped. This would be a harder attack to put forth when meticulously keyed authentication is in use. o BFD works on microsecond timers. When malicious packets are sent at short intervals, with the authentication bit set, it can cause a DoS attack. o BFD allows a mode called the echo mode. Echo packets are not defined in the BFD specification, though they can keep the BFD session up. There are no guidelines on the properties of the echo packets beyond the choice of the source and destination addresses. While the BFD specification recommends applying security mechanisms to prevent spoofing of these packets, there are no guidelines on what type of mechanisms are appropriate. Any security issues in the echo mode will directly affect the BFD protocol and session states, and hence the network stability. The potential effects and remedies as understood today are somewhat limited, however. For instance, any replay attacks would be indistinguishable from normal forwarding of the tested router. An attack would still cause a faulty link to be believed to be up, but there is little that can be done about it. However, if the echo packets are guessable, it may be possible to spoof from an external source and cause BFD to believe that a one-way link is really bidirectional. As a result, it is important that the echo packets contain random material that is also checked upon reception. o BFD packets can be sent at millisecond intervals (the protocol uses timers at microsecond intervals). When using authentication, this can cause frequent sequence number wrap-around as a 32-bit sequence number is used, thus considerably reducing the security of the authentication algorithms. o Recently, limitations in collision-resistance properties of the MD5 and SHA-1 hash functions have been discovered; [RFC4270] summarizes the discoveries. There have been attacks against the use of MD5 as a hash; while these attacks do not directly apply to the use of HMAC-MD5 and keyed SHA-1 in BFD, it is prudent to have other options available. Such attacks do not mean that BFD using SHA-1 for authentication is at risk. However, it does mean that SHA-1 should be replaced as soon as possible and should not be used for new applications.
It should be noted that if SHA-1 is used in the Hashed Message Authentication Code (HMAC) [RFC2104] construction, then collision attacks currently known against SHA-1 do not apply. The new attacks on SHA-1 have no impact on the security of HMAC-SHA-1. There are already proposals [GenBFDAuth] that add support for HMAC with the SHA-1 and SHA-2 family of hash functions for BFD. 8. Security Considerations This document outlines security issues arising from the current methodology for manual keying of various routing protocols. No specific changes to routing protocols are proposed in this document; likewise, no new security requirements result. 9. Acknowledgements We would like to acknowledge Sam Hartman, Ran Atkinson, Stephen Kent and Brian Weis for their initial comments on this document. Thanks to Merike Kaeo and Alfred Hoenes for reviewing many sections of the document and providing lot of useful comments. 10. References 10.1. Normative References [RFC0793] Postel, J., "Transmission Control Protocol", STD 7, RFC 793, September 1981. [RFC1195] Callon, R., "Use of OSI IS-IS for routing in TCP/IP and dual environments", RFC 1195, December 1990. [RFC2328] Moy, J., "OSPF Version 2", STD 54, RFC 2328, April 1998. [RFC2385] Heffernan, A., "Protection of BGP Sessions via the TCP MD5 Signature Option", RFC 2385, August 1998. [RFC2453] Malkin, G., "RIP Version 2", STD 56, RFC 2453, November 1998. [RFC4271] Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A Border Gateway Protocol 4 (BGP-4)", RFC 4271, January 2006. [RFC4302] Kent, S., "IP Authentication Header", RFC 4302, December 2005. Kent, S., "IP Authentication Header",
[RFC4303] Kent, S., "IP Encapsulating Security Payload (ESP)", RFC 4303, December 2005. [RFC4552] Gupta, M. and N. Melam, "Authentication/Confidentiality for OSPFv3", RFC 4552, June 2006. [RFC4822] Atkinson, R. and M. Fanto, "RIPv2 Cryptographic Authentication", RFC 4822, February 2007. [RFC5340] Coltun, R., Ferguson, D., Moy, J., and A. Lindem, "OSPF for IPv6", RFC 5340, July 2008. [RFC5304] Li, T. and R. Atkinson, "IS-IS Cryptographic Authentication", RFC 5304, October 2008. [RFC5310] Bhatia, M., Manral, V., Li, T., Atkinson, R., White, R., and M. Fanto, "IS-IS Generic Cryptographic Authentication", RFC 5310, February 2009. 10.2. Informative References [Arjen05] Arjen K. Lenstra, "Further progress in Hashing cryptanalysis", Lucent Bell Laboratories, February 26, 2005. [Conv01] Convery, et al., "BGP Vulnerability Testing: Separating Fact from FUD v1.00", NANOG 28, pp. 1-61, June 2003. [Crypto2004] Xiaoyun Wang, Xuejia Lai, Dengguo Feng, and Hongbo Yu, "Collisions for hash functions MD4, MD5, HAVAL-128, and RIPEMD", Crypto 2004 Rump Session. [Dobb96a] Dobbertin, H., "Cryptanalysis of MD5 Compress", Technical Report, 2 May 1996. (Presented at the Rump Session of EuroCrypt 1996.) [Dobb96b] Dobbertin, H., "The Status of MD5 After a Recent Attack", CryptoBytes, Vol. 2, No. 2, Summer 1996. [GenBFDAuth] Bhatia, M. and V. Manral, "BFD Generic Cryptographic Authentication", Work in Progress, June 2010. [NISTHmacSHA] "NIST's Policy on Hash Functions", 2006, http://csrc.nist.gov/groups/ST/hash/policy.html.
[Philip01] Philip Hawkes, Michael Paddon, and Gregory G. Rose, "On Corrective Patterns for the SHA-2 Family", IACR ePrint Archive, 2004, http://eprint.iacr.org/2004/207. [Prav01] Praveen Gauravaram, et al., "Collision Attacks on MD5 and SHA-1: Is this the 'Sword of Domocles' for Electronic Commerce?", Information Security Institue (ISI), Queensland University of Technology (QUT), Australia. [Prav02] Praveen Gauravaram, et al., "Some thoughts on Collision Attacks in the Hash Functions Md5, SHA-0 and SHA-1", Information Security Institue (ISI), Queensland University of Technology (QUT), Australia. [RFC1058] Hedrick, C., "Routing Information Protocol", RFC 1058, June 1988. [RFC1321] Rivest, R., "The MD5 Message-Digest Algorithm", RFC 1321, April 1992. [RFC1388] Malkin, G., "RIP Version 2 Carrying Additional Information", RFC 1388, January 1993. [RFC1723] Malkin, G., "RIP Version 2 - Carrying Additional Information", RFC 1723, November 1994. [RFC1923] Halpern, J. and S. Bradner, "RIPv1 Applicability Statement for Historic Status", RFC 1923, March 1996. [RFC2082] Baker, F. and R. Atkinson, "RIP-2 MD5 Authentication", RFC 2082, January 1997. [RFC2104] Krawczyk, H., Bellare, M., and R. Canetti, "HMAC: Keyed-Hashing for Message Authentication", RFC 2104, February 1997. [RFC2154] Murphy, S., Badger, M., and B. Wellington, "OSPF with Digital Signatures", RFC 2154, June 1997. [RFC2410] Glenn, R. and S. Kent, "The NULL Encryption Algorithm and Its Use With IPsec", RFC 2410, November 1998. [RFC3562] Leech, M., "Key Management Considerations for the TCP MD5 Signature Option", RFC 3562, July 2003.
[RFC4270] Hoffman, P. and B. Schneier, "Attacks on Cryptographic Hashes in Internet Protocols", RFC 4270, November 2005. [RFC4306] Kaufman, C., Ed., "Internet Key Exchange (IKEv2) Protocol", RFC 4306, December 2005. [RFC5709] Bhatia, M., Manral, V., Fanto, M., White, R., Barnes, M., Li, T., and R. Atkinson, "OSPFv2 HMAC-SHA Cryptographic Authentication", RFC 5709, October 2009. [RFC5880] Katz, D. and D. Ward, "Bidirectional Forwarding Detection (BFD)", RFC 5880, June 2010. [RFC5881] Katz, D. and D. Ward, "Bidirectional Forwarding Detection (BFD) for IPv4 and IPv6 (Single Hop)", RFC 5881, June 2010. [RFC5883] Katz, D. and D. Ward, "Bidirectional Forwarding Detection (BFD) for Multihop Paths", RFC 5883, June 2010. [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP Authentication Option", RFC 5925, June 2010. [RR07] Rechberger, C. and V. Rijmen, "On Authentication with HMAC and Non-random Properties", Financial Cryptography and Data Security, Lecture Notes in Computer Science, Volume 4886/2008, Springer-Verlag, Berlin, December 2007. [RR08] Rechberger, C. and V. Rijmen, "New Results on NMAC/HMAC when Instantiated with Popular Hash Functions", Journal of Universal Computer Science, Volume 14, Number 3, pp. 347-376, 1 February 2008. [TcpWindow] Watson, P., "Slipping in the Window: TCP Reset attacks", Presentation at 2004 CanSecWest, http://cansecwest.com/csw04archive.html. [Wang04] Wang, X., et al., "Collisions for Hash Functions MD4, MD5, HAVAL-128 and RIPEMD", August 2004, IACR ePrint Archive, http://eprint.iacr.org/2004/199.
[Wang05] Wang, X., et al., "Finding Collisions in the Full SHA-1", Proceedings of Crypto 2005, Lecture Notes in Computer Science, Volume 3621, pp. 17-36, Springer- Verlag, Berlin, August 31, 2005. 11. Contributor's Address Sue Hares NextHop USA EMail: email@example.com Authors' Addresses Vishwas Manral IP Infusion, Inc. 1188 E. Arques Ave. Sunnyvale, CA 94085 EMail: firstname.lastname@example.org Manav Bhatia Alcatel-Lucent Bangalore India EMail: email@example.com Joel P. Jaeggli Nokia Inc. EMail: firstname.lastname@example.org Russ White Cisco Systems RTP North Carolina USA EMail: email@example.com