| Network Working Group | M. Bhatia |
| Internet-Draft | Alcatel-Lucent |
| Intended status: Standards Track | V. Manral |
| Expires: October 19, 2014 | Hewlett-Packard Co. |
| D. Zhang | |
| Huawei | |
| M. Jethanandani | |
| Ciena Corporation | |
| April 17, 2014 |
BFD Generic Cryptographic Authentication
draft-ietf-bfd-generic-crypto-auth-06
This document proposes an extension to Bidirectional Forwarding Detection (BFD) to allow the use of arbitrary cryptographic authentication algorithms in addition to the already-documented authentication schemes described in the base specification. This document adds the basic infrastructure that is required for supporting algorithm and key agility for BFD.
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 [RFC2119].
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 http://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 October 19, 2014.
Copyright (c) 2014 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.
The base specification of Bidirectional Forwarding Detection [RFC5880] defines five authentication schemes: Simple Password, Keyed MD5, Meticulous Keyed MD5, Keyed SHA-1, and Meticulous SHA-1. In Simple Password, passwords are transferred in plain text. An attacker with physical access to the network can easily eavesdrop on the password and compromise the security of the BFD packet exchanges. In Keyed MD5 and Meticulous Keyed MD5, the BFD devices on the both sides of a BFD session share a secret key which is used to generate a keyed MD5 digest for each packet, and a monotonically increasing sequence number scheme is used to prevent replay attacks. Keyed SHA-1 and Meticulous SHA-1 modes are similar to MD5, and it uses SHA-1 instead of MD5 to generate a digest for each packet.
The security strength of the cryptographic algorithms adopted in the authentication schemes are relatively weak. Both the MD5 algorithm and the SHA-1 algorithm are known to be vulnerable to collision attacks. In MD5-attack [MD5-attack] and Dobb96a [Dobb96a], Dobb96b [Dobb96b], several methods of generating hash collisions for some applications of MD5 are proposed. Similar security vulnerabilities of SHA-1 are introduced in SHA-1-attack1 [SHA-1-attack1] and SHA-1-attack2 [SHA-1-attack2]. It is therefore desired that BFD must support newer algorithms that have not yet been broken. Additionally, the transition mechanism from one algorithm to the other must be seamless.
The other issue with the existing authentication schemes is the vulnerability to replay attacks. In non-meticulous authentication schemes, sequence numbers are only increased occasionally. This behavior can be taken advantage of by an attacker to perform intra-session replay attacks. In meticulous authentication schemes, sequence numbers are required to monotonically increase with each successive packet, which eliminates the possibility of intra-session replay attacks.
BFD session timers are often defined with the granularity of microseconds. Although in practice BFD devices send packets at millisecond intervals, they can potentially send packets at a much higher rate. Since the cryptographic sequence number space is only 32 bits, when using Meticulous Authentication, a sequence number used in a BFD session can reach its maximum value and roll over within a short period. For instance, if the value of a sequence number is increased by one every millisecond, then it will reach its maximum in less than 8 weeks. This can potentially be exploited to launch inter-session replay attacks.
In order to address the issues mentioned above, this document proposes two new authentication types that can be used to secure the BFD packets. The two authentication types are - Cryptographic Authentication (CRYPTO_AUTH) and Meticulous Cryptographic Authentication (MET_ CRYPTO_AUTH). Unlike earlier authentication types that were defined in BFD, the proposed authentication types are not tied to any particular authentication algorithm or construct. These can use different authentication algorithms and constructs like MD5, SHA-1, SHA-2, HMAC-SHA1, HMAC-SHA2, etc. to provide authentication and data integrity protection for BFD control packets.
The packet replay mechanism has also been enhanced to improve its capability in handling inter and intra-session replay attacks.
It should be noted that this document attempts to fix the security issues raised by the manual key management procedure that currently exists within BFD, as part of the Phase One described in KARP Design Guidelines [RFC6518]. Therefore, only the pre-shared keys is considered in this document. However, the solution described in this document is generic and does not preclude the possibility of supporting keys derived from an automated key management protocol.
The BFD protocol does not include an in-band mechanism to create or manage BFD Security Associations (BFD SA). A BFD SA contains a set of shared parameters between any two legitimate BFD devices.
The parameters associated with a BFD SA are listed as follows:
In the proposed authentication extension, an optional authentication section (Generic Authentication Section) and two authentication types (Generic Cryptographic Authentication and Generic Meticulous Cryptographic Authentication) are specified.
The Authentication section is only present in a BFD packet if the Authentication Present (A) bit is set in the packet header. The Auth Type in the Authentication section is set to TBD1 when Generic Cryptographic Authentication is in use, while it is set to TBD2 when Generic Meticulous Cryptographic Authentication is in use.
Both the authentication types use a monotonically increasing sequence number to protect the BFD session against reply attacks. The only difference between the two types is that the sequence number is occasionally incremented in the Cryptographic Authentication mode, as against the Meticulous Cryptographic Authentication mode, where it is incremented on every packet.
As a result of this, in the Cryptographic Authentication scheme, a replay attack is possible till the next sequence number is sent out.
A new authentication type, TBD1 or TBD2, indicating the generic cryptographic authentication mechanism in use, is inserted in the first octet of Authentication Section of the BFD control packet.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Auth Type | Auth Len | Auth Key ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Sequence Number (High Order 32 Bits) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Sequence Number (Low Order 32 Bits) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Authentication Data (Variable) | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
For a BFD packet, if the Authentication Present (A) bit is set in the header and the Authentication Type field is TBD1 (Generic Cryptographic Authentication) or TBD2 (Generic Meticulous Cryptographic Authentication), the Authentication Section has the following format:
Before a BFD device sends a BFD packet out, the device needs to select an appropriate BFD SA from its local key database if a keyed digest for the packet is required. If no appropriate SA is available, the BFD packet MUST be discarded.
If an appropriate SA is available, the device then derives the key and the associated authentication algorithm from the SA.
The device sets the Authentication Present (A) bit in the packet header.
The device MUST fill the Auth Type, the Auth Len fields and the Sequence Number field to bfd.XmitAuthSeq before the authentication data is computed.
The Auth Len field in the Authentication section is set as per the authentication algorithm that is being used.
The Key ID field is filled.
The computation of the digest is performed. The computing process can be various when different algorithms are adopted and is out of the scope of this document.
The generated digest is placed in the Authentication Data field.
When a BFD Control packet is received, the following procedure MUST be followed, in the order specified.
If the Authentication Present (A) bit is set in the packet header and the Auth Type is TBD1 or TBD2, the receiver is to find an appropriate BFD SA in its local key table to process the packet. The BFD SA is identified by the Key ID field in the Authentication Section of the incoming BFD packet.
If the Auth Key ID field does not match the ID of any configured authentication key or the associated key is not in its valid period, the received packet MUST be discarded.
If bfd.AuthSeqKnown is 1, examine the Sequence Number field. For Cryptographic Authentication, if the Sequence Number lies outside of the range of bfd.RcvAuthSeq to bfd.RcvAuthSeq+(3*Detect Mult) inclusive (when treated as an unsigned 64 bit circular number space), the received packet MUST be discarded. For Meticulous Cryptographic Authentication, if the Sequence Number lies outside of the range of bfd.RcvAuthSeq+1 to bfd.RcvAuthSeq+(3*Detect Mult) inclusive (when treated as an unsigned 64 bit circular number space, the received packet MUST be discarded.
The device then prepares for generating a digest of the packet. First of all, the authentication data in the Authentication Value field needs to be saved somewhere else. Then the Authentication Value field is set with a pre-specified value (which may be various in different security algorithms) according the authentication algorithm indicated in the SA. After this, the device starts performing the digest generating operations. The work of defining actual digest generating operations is out of the scope of this document.
The calculated data is compared with the received authentication data in the packet and the packet MUST be discarded if the two do not match. In such a case, an error event SHOULD be logged.
An implementation MAY have a transition mode where it includes CRYPTO_AUTH or the MET_CRYPTO_AUTH information in the packets but does not verify this information. This is provided as a transition aid for networks in the process of migrating to the new CRYPTO_AUTH and MET_CRYPTO_AUTH based authentication schemes.
In [I-D.ietf-karp-crypto-key-table], a conceptual key database called "key table" is introduced. A key table is located in the middle of key management protocols and security protocols so that a security protocol can derive long-term keys from the key table but does not have to know the details of key management. This section describes how the proposed security solution selects long-lived keys from key tables [I-D.ietf-karp-crypto-key-table].
Assume that a device R1 tries to send a unicast BFD packet from its interface I1 to the interface R2 of a remote device R2 at time T. Because the key should be shared by the by both R1 and R2 to protect the communication between I1 and I2, R1 needs to provide a protocol ("BFD"), an interface identifier (I1) and a peer identifier (R2) into the key selection function. Any key that satisfies the following conditions may be selected:
After a set of keys are provided, a BFD implementation should support selection of keys based on algorithm preference.
Upon reception of a BFD packet from R1, R2 provides the protocol ("BFD"), the peer identifier (R1), the key identifier derived from the incoming packet (L), and the interface (I2) to the key table. Any key that satisfies the following conditions may be selected:
As described in Section 1, if the BFD packets in a session are transferred with a high frequency, a 32-bit sequence number may reach its maximum and have to roll back before the session finishes. A attacker thus can replay the packets intercepted before the sequence number wrapped without being detected. To address this problem, the length of the sequence number in the proposed authentication section has been extended to 64 bits. After the extension, the sequence number space of a device will not be exhausted for half of a million years even if the device sends out a BFD packet in every micro-second. Therefore, the replay attack risks caused by the limited sequence number space can be largely addressed. However, in Generic Cryptographic Authentication, the sequence number is only required to increase occasionally. Therefore, a replayed packet may be regarded as a legal one until the packet with a larger sequence number is received. This type of intra-session replay attack cannot be addressed only by extending the length of sequence numbers.
An anti-replay solution for BFD also needs to consider the scenarios where a BFD device loses its prior sequence number state (e.g., system crash, loss of power). In such cases, a BFD device has to re-initialize its sequence number. Otherwise, an attacker may be able to replay a previously intercepted without being detected.
To address this problem, in the proposed solution, the most significant 32-bit value of the sequence number is used to contain a boot count, and the remainder 32-bit value is used as an ordinary 32-bit monotonically increasing sequence number. In Generic Cryptographic Authentication, the remainder 32-bit value is required to increase occasionally, while in Generic Meticulous Cryptographic Authentication, the lower order 32-bit sequence number MUST be incremented for every BFD packet sent by a BFD device. The BFD implementations are required to retain the boot count in non-volatile storage for the deployment life the BFD device. The boot count increases each time when the BFD device loses its prior sequence number state. The SNMPv3 snmpEngineBoots variable [RFC4222] MAY be used for this purpose. However, maintaining a separate boot count solely for BFD sequence numbers has the advantage of decoupling SNMP re-initialization and BFD re-initialization. Also, in the rare event that the lower order 32- bit sequence number wraps, the boot count can be incremented to preserve the strictly increasing property of the aggregate sequence number. Hence, a separate BFD boot count is RECOMMENDED.
IANA is requested to assign two authentication types from the "BFD Authentication Types" sub-registry within the "Bidirectional Forwarding Detection (BFD) Parameters" registry.
| Address | BFD Authentication Type Name | Reference |
|---|---|---|
| TBD1 | Cryptographic Authentication | This document |
| TBD2 | Meticulous Cryptographic Authentication | This document |
The proposed sequence number extension offers most of the benefits of more complicated mechanisms involving challenges. There are, however, a couple drawbacks to this approach.
First, it requires the BFD implementation to be able to save its boot count in non-volatile storage. If the non-volatile storage is ever repaired or upgraded such that the contents are lost or the BFD device is replaced with a model, the keys MUST be changed to prevent replay attacks.
Second, if a device is taken out of service completely (either intentionally or due to a persistent failure), the potential exists for re-establishment of a BFD adjacency by replaying the entire BFD session establishment. This scenario is however, extremely unlikely and can be easily avoided. For instance, after recovering from a system failure, a BFD device has to re-establish BFD sessions. At this stage, if the device randomly selects its discriminators to identify new BFD sessions, the possibility of re-establishing a BFD session by replaying the entire BFD session establishment will be eliminated. For the implementations in which discriminators are not randomly selected, this issue can be largely mitigated by integrating the boot count of the remote BFD router in the generation of the authentication data for outgoing BFD packets. Of course, this attack could also be thwarted by changing the relevant manual keys.
There is a transition mode suggested where devices can ignore the CRYPTO_AUTH or the MET_CRYPTO_AUTH information carried in the packets. The operator must ensure that this mode is only used when migrating to the new CRYPTO_AUTH/MET_CRYPTO_AUTH based authentication scheme as this leaves the device vulnerable to an attack.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. |
| [RFC5880] | Katz, D. and D. Ward, "Bidirectional Forwarding Detection (BFD)", RFC 5880, June 2010. |