Network Working Group R. Atkinson Request for Comments: 1826 Naval Research Laboratory Category: Standards Track August 1995 IP Authentication Header Status of this Memo This document specifies an Internet standards track protocol for the Internet community, and requests discussion and suggestions for improvements. Please refer to the current edition of the "Internet Official Protocol Standards" (STD 1) for the standardization state and status of this protocol. Distribution of this memo is unlimited. ABSTRACT This document describes a mechanism for providing cryptographic authentication for IPv4 and IPv6 datagrams. An Authentication Header (AH) is normally inserted after an IP header and before the other information being authenticated. 1. INTRODUCTION The Authentication Header is a mechanism for providing strong integrity and authentication for IP datagrams. It might also provide non-repudiation, depending on which cryptographic algorithm is used and how keying is performed. For example, use of an asymmetric digital signature algorithm, such as RSA, could provide non- repudiation. Confidentiality, and protection from traffic analysis are not provided by the Authentication Header. Users desiring confidentiality should consider using the IP Encapsulating Security Protocol (ESP) either in lieu of or in conjunction with the Authentication Header [Atk95b]. This document assumes the reader has previously read the related IP Security Architecture document which defines the overall security architecture for IP and provides important background information for this specification [Atk95a]. 1.1 Overview The IP Authentication Header seeks to provide security by adding authentication information to an IP datagram. This authentication information is calculated using all of the fields in the IP datagram (including not only the IP Header but also other headers and the user data) which do not change in transit. Fields or options which need to change in transit (e.g., "hop count", "time to live", "ident", Atkinson Standards Track [Page 1] RFC 1826 IP Authentication Header August 1995 "fragment offset", or "routing pointer") are considered to be zero for the calculation of the authentication data. This provides significantly more security than is currently present in IPv4 and might be sufficient for the needs of many users. Use of this specification will increase the IP protocol processing costs in participating end systems and will also increase the communications latency. The increased latency is primarily due to the calculation of the authentication data by the sender and the calculation and comparison of the authentication data by the receiver for each IP datagram containing an Authentication Header. The impact will vary with authentication algorithm used and other factors. In order for the Authentication Header to work properly without changing the entire Internet infrastructure, the authentication data is carried in its own payload. Systems that aren't participating in the authentication MAY ignore the Authentication Data. When used with IPv6, the Authentication Header is normally placed after the Fragmentation and End-to-End headers and before the ESP and transport-layer headers. The information in the other IP headers is used to route the datagram from origin to destination. When used with IPv4, the Authentication Header immediately follows an IPv4 header. If a symmetric authentication algorithm is used and intermediate authentication is desired, then the nodes performing such intermediate authentication would need to be provided with the appropriate keys. Possession of those keys would permit any one of those systems to forge traffic claiming to be from the legitimate sender to the legitimate receiver or to modify the contents of otherwise legitimate traffic. In some environments such intermediate authentication might be desirable [BCCH94]. If an asymmetric authentication algorithm is used and the routers are aware of the appropriate public keys and authentication algorithm, then the routers possessing the authentication public key could authenticate the traffic being handled without being able to forge or modify otherwise legitimate traffic. Also, Path MTU Discovery MUST be used when intermediate authentication of the Authentication Header is desired and IPv4 is in use because with this method it is not possible to authenticate a fragment of a packet [MD90] [Kno93]. Atkinson Standards Track [Page 2] RFC 1826 IP Authentication Header August 1995 1.2 Requirements Terminology In this document, the words that are used to define the significance of each particular requirement are usually capitalised. These words are: - MUST This word or the adjective "REQUIRED" means that the item is an absolute requirement of the specification. - SHOULD This word or the adjective "RECOMMENDED" means that there might exist valid reasons in particular circumstances to ignore this item, but the full implications should be understood and the case carefully weighed before taking a different course. - MAY This word or the adjective "OPTIONAL" means that this item is truly optional. One vendor might choose to include the item because a particular marketplace requires it or because it enhances the product, for example; another vendor may omit the same item. 2. KEY MANAGEMENT Key management is an important part of the IP security architecture. However, it is not integrated with this specification because of a long history in the public literature of subtle flaws in key management algorithms and protocols. The IP Authentication Header tries to decouple the key management mechanisms from the security protocol mechanisms. The only coupling between the key management protocol and the security protocol is with the Security Parameters Index (SPI), which is described in more detail below. This decoupling permits several different key management mechanisms to be used. More importantly, it permits the key management protocol to be changed or corrected without unduly impacting the security protocol implementations. The key management mechanism is used to negotiate a number of parameters for each "Security Association", including not only the keys but also other information (e.g., the authentication algorithm and mode) used by the communicating parties. The key management mechanism creates and maintains a logical table containing the several parameters for each current security association. An implementation of the IP Authentication Header will need to read that Atkinson Standards Track [Page 3] RFC 1826 IP Authentication Header August 1995 logical table of security parameters to determine how to process each datagram containing an Authentication Header (e.g., to determine which algorithm/mode and key to use in authentication). Security Associations are unidirectional. A bidirectional communications session will normally have one Security Association in each direction. For example, when a TCP session exists between two systems A and B, there will normally be one Security Association from A to B and a separate second Security Assocation from B to A. The receiver assigns the SPI value to the the Security Association with that sender. The other parameters of the Security Association are determined in a manner specified by the key management mechanism. Section 4 of this document describes in detail the process of selecting a Security Association for an outgoing packet and identifying the Security Assocation for an incoming packet. The IP Security Architecture document describes key management in detail. It includes specification of the key management requirements for this protocol, and is incorporated here by reference [Atk95a]. 3. AUTHENTICATION HEADER SYNTAX The Authentication Header (AH) may appear after any other headers which are examined at each hop, and before any other headers which are not examined at an intermediate hop. The IPv4 or IPv6 header immediately preceding the Authentication Header will contain the value 51 in its Next Header (or Protocol) field [STD-2]. Example high-level diagrams of IP datagrams with the Authentication Header follow. +------------+-------------------+------------+-------+---------------+ | IPv6 Header| Hop-by-Hop/Routing| Auth Header| Others| Upper Protocol| +------------+-------------------+------------+-------+---------------+ Figure 1: IPv6 Example Atkinson Standards Track [Page 4] RFC 1826 IP Authentication Header August 1995 When used with IPv6, the Authentication Header normally appears after the IPv6 Hop-by-Hop Header and before the IPv6 Destination Options. +-------------+--------------+-------------------------------+ | IPv4 Header | Auth Header | Upper Protocol (e.g. TCP, UDP)| +-------------+--------------+-------------------------------+ Figure 2: IPv4 Example When used with IPv4, the Authentication Header normally follows the main IPv4 header. 3.1 Authentication Header Syntax The authentication data is the output of the authentication algorithm calculated over the the entire IP datagram as described in more detail later in this document. The authentication calculation must treat the Authentication Data field itself and all fields that are normally modified in transit (e.g., TTL or Hop Limit) as if those fields contained all zeros. All other Authentication Header fields are included in the authentication calculation normally. The IP Authentication Header has the following syntax: +---------------+---------------+---------------+---------------+ | Next Header | Length | RESERVED | +---------------+---------------+---------------+---------------+ | Security Parameters Index | +---------------+---------------+---------------+---------------+ | | + Authentication Data (variable number of 32-bit words) | | | +---------------+---------------+---------------+---------------+ 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 Figure 3: Authentication Header syntax Atkinson Standards Track [Page 5] RFC 1826 IP Authentication Header August 1995 3.2 Fields of the Authentication Header NEXT HEADER 8 bits wide. Identifies the next payload after the Authentication Payload. This values in this field are the set of IP Protocol Numbers as defined in the most recent RFC from the Internet Assigned Numbers Authority (IANA) describing "Assigned Numbers" [STD-2]. PAYLOAD LENGTH 8 bits wide. The length of the Authentication Data field in 32- bit words. Minimum value is 0 words, which is only used in the degenerate case of a "null" authentication algorithm. RESERVED 16 bits wide. Reserved for future use. MUST be set to all zeros when sent. The value is included in the Authentication Data calculation, but is otherwise ignored by the recipient. SECURITY PARAMETERS INDEX (SPI) A 32-bit pseudo-random value identifying the security association for this datagram. The Security Parameters Index value 0 is reserved to indicate that "no security association exists". The set of Security Parameters Index values in the range 1 through 255 are reserved to the Internet Assigned Numbers Authority (IANA) for future use. A reserved SPI value will not normally be assigned by IANA unless the use of that particular assigned SPI value is openly specified in an RFC. AUTHENTICATION DATA This length of this field is variable, but is always an integral number of 32-bit words. Many implementations require padding to other alignments, such as 64-bits, in order to improve performance. All implementations MUST support such padding, which is specified by the Destination on a per SPI basis. The value of the padding field is arbitrarily selected by the sender and is included in the Authentication Data calculation. An implementation will normally use the combination of Destination Address and SPI to locate the Security Association which specifies the field's size and use. The field retains the same format for all datagrams of any given SPI and Destination Address pair. Atkinson Standards Track [Page 6] RFC 1826 IP Authentication Header August 1995 The Authentication Data fills the field beginning immediately after the SPI field. If the field is longer than necessary to store the actual authentication data, then the unused bit positions are filled with unspecified, implementation-dependent values. Refer to each Authentication Transform specification for more information regarding the contents of this field. 3.3 Sensitivity Labeling As is discussed in greater detail in the IP Security Architecture document, IPv6 will normally use implicit Security Labels rather than the explicit labels that are currently used with IPv4 [Ken91] [Atk95a]. In some situations, users MAY choose to carry explicit labels (for example, IPSO labels as defined by RFC-1108 might be used with IPv4) in addition to using the implicit labels provided by the Authentication Header. Explicit label options could be defined for use with IPv6 (e.g., using the IPv6 end-to-end options header or the IPv6 hop-by-hop options header). Implementations MAY support explicit labels in addition to implicit labels, but implementations are not required to support explicit labels. If explicit labels are in use, then the explicit label MUST be included in the authentication calculation. 4. CALCULATION OF THE AUTHENTICATION DATA The authentication data carried by the IP Authentication Header is usually calculated using a message digest algorithm (for example, MD5) either encrypting that message digest or keying the message digest directly [Riv92]. Only algorithms that are believed to be cryptographically strong one-way functions should be used with the IP Authentication Header. Because conventional checksums (e.g., CRC-16) are not cryptographically strong, they MUST NOT be used with the Authentication Header. When processing an outgoing IP packet for Authentication, the first step is for the sending system to locate the appropriate Security Association. All Security Associations are unidirectional. The selection of the appropriate Security Association for an outgoing IP packet is based at least upon the sending userid and the Destination Address. When host-oriented keying is in use, all sending userids will share the same Security Association to a given destination. When user-oriented keying is in use, then different users or possibly even different applications of the same user might use different Security Associations. The Security Association selected will Atkinson Standards Track [Page 7] RFC 1826 IP Authentication Header August 1995 indicate which algorithm, algorithm mode, key, and other security properties apply to the outgoing packet. Fields which NECESSARILY are modified during transit from the sender to the receiver (e.g., TTL and HEADER CHECKSUM for IPv4 or Hop Limit for IPv6) and whose value at the receiver are not known with certainty by the sender are included in the authentication data calculation but are processed specially. For these fields which are modified during transit, the value carried in the IP packet is replaced by the value zero for the purpose of the authentication calculation. By replacing the field's value with zero rather than omitting these fields, alignment is preserved for the authentication calculation. The sender MUST compute the authentication over the packet as that packet will appear at the receiver. This requirement is placed in order to allow for future IP optional headers which the receiver might not know about but the sender necessarily knows about if it is including such options in the packet. This also permits the authentication of data that will vary in transit but whose value at the final receiver is known with certainty by the sender in advance. The sender places the calculated message digest algorithm output into the Authentication Data field within the Authentication Header. For purposes of Authentication Data computation, the Authentication Data field is considered to be filled with zeros. The IPv4 "TIME TO LIVE" and "HEADER CHECKSUM" fields are the only fields in the IPv4 base header that are handled specially for the Authentication Data calculation. Reassembly of fragmented packets occurs PRIOR to processing by the local IP Authentication Header implementation. The "more" bit is of course cleared upon reassembly. Hence, no other fields in the IPv4 header will vary in transit from the perspective of the IP Authentication Header implementation. The "TIME TO LIVE" and "HEADER CHECKSUM" fields of the IPv4 base header MUST be set to all zeros for the Authentication Data calculation. All other IPv4 base header fields are processed normally with their actual contents. Because IPv4 packets are subject to intermediate fragmentation in routers, it is important that the reassembly of IPv4 packets be performed prior to the Authentication Header processing. IPv4 Implementations SHOULD use Path MTU Discovery when the IP Authentication Header is being used [MD90]. For IPv4, not all options are openly specified in a RFC, so it is not possible to enumerate in this document all of the options that might normally be modified during transit. The IP Security Option (IPSO) MUST be included in the Authentication Data calculation whenever that option is present in an IP datagram [Ken91]. If a receiving system does not recognise an IPv4 option that is present in the packet, that option Atkinson Standards Track [Page 8] RFC 1826 IP Authentication Header August 1995 is included in the Authentication Data calculation. This means that any IPv4 packet containing an IPv4 option that changes during transit in a manner not predictable by the sender and which IPv4 option is unrecognised by the receiver will fail the authentication check and consequently be dropped by the receiver. The IPv6 "HOP LIMIT" field is the only field in the IPv6 base header that is handled specially for Authentication Data calculation. The value of the HOP LIMIT field is zero for the purpose of Authentication Data calculation. All other fields in the base IPv6 header MUST be included in the Authentication Data calculation using the normal procedures for calculating the Authentication Data. All IPv6 "OPTION TYPE" values contain a bit which MUST be used to determine whether that option data will be included in the Authentication Data calculation. This bit is the third-highest-order bit of the IPv6 OPTION TYPE field. If this bit is set to zero, then the corresponding option is included in the Authentication Data calculation. If this bit is set to one, then the corresponding option is replaced by all zero bits of the same length as the option for the purpose of the Authentication Data calculation. The IPv6 Routing Header "Type 0" will rearrange the address fields within the packet during transit from source to destination. However, this is not a problem because the contents of the packet as it will appear at the receiver are known to the sender and to all intermediate hops. Hence, the IPv6 Routing Header "Type 0" is included in the Authentication Data calculation using the normal procedure. Upon receipt of a packet containing an IP Authentication Header, the receiver first uses the Destination Address and SPI value to locate the correct Security Association. The receiver then independently verifies that the Authentication Data field and the received data packet are consistent. Again, the Authentication Data field is assumed to be zero for the sole purpose of making the authentication computation. Exactly how this is accomplished is algorithm dependent. If the processing of the authentication algorithm indicates the datagram is valid, then it is accepted. If the algorithm determines that the data and the Authentication Header do not match, then the receiver SHOULD discard the received IP datagram as invalid and MUST record the authentication failure in the system log or audit log. If such a failure occurs, the recorded log data MUST include the SPI value, date/time received, clear-text Sending Address, clear-text Destination Address, and (if it exists) the clear-text Flow ID. The log data MAY also include other information about the failed packet. Atkinson Standards Track [Page 9] RFC 1826 IP Authentication Header August 1995 5. CONFORMANCE REQUIREMENTS Implementations that claim conformance or compliance with this specification MUST fully implement the header described here, MUST support manual key distribution for use with this option, MUST comply with all requirements of the "Security Architecture for the Internet Protocol" [Atk95a], and MUST support the use of keyed MD5 as described in the companion document entitled "IP Authentication using Keyed MD5" [MS95]. Implementations MAY also implement other authentication algorithms. Implementors should consult the most recent version of the "IAB Official Standards" RFC for further guidance on the status of this document. 6. SECURITY CONSIDERATIONS This entire RFC discusses an authentication mechanism for IP. This mechanism is not a panacea to the several security issues in any internetwork, however it does provide a component useful in building a secure internetwork. Users need to understand that the quality of the security provided by this specification depends completely on the strength of whichever cryptographic algorithm has been implemented, the strength of the key being used, the correctness of that algorithm's implementation, upon the security of the key management mechanism and its implementation, and upon the correctness of the IP Authentication Header and IP implementations in all of the participating systems. If any of these assumptions do not hold, then little or no real security will be provided to the user. Implementors are encouraged to use high assurance methods to develop all of the security relevant parts of their products. Users interested in confidentiality should consider using the IP Encapsulating Security Payload (ESP) instead of or in conjunction with this specification [Atk95b]. Users seeking protection from traffic analysis might consider the use of appropriate link encryption. Description and specification of link encryption is outside the scope of this note [VK83]. Users interested in combining the IP Authentication Header with the IP Encapsulating Security Payload should consult the IP Encapsulating Security Payload specification for details. One particular issue is that in some cases a packet which causes an error to be reported back via ICMP might be so large as not to entirely fit within the ICMP message returned. In such cases, it might not be possible for the receiver of the ICMP message to independently authenticate the portion of the returned message. This could mean that the host receiving such an ICMP message would either Atkinson Standards Track [Page 10] RFC 1826 IP Authentication Header August 1995 trust an unauthenticated ICMP message, which might in turn create some security problem, or not trust and hence not react appropriately to some legitimate ICMP message that should have been reacted to. It is not clear that this issue can be fully resolved in the presence of packets that are the same size as or larger than the minimum IP MTU. Similar complications arise if an encrypted packet causes an ICMP error message to be sent and that packet is truncated. Active attacks are now widely known to exist in the Internet [CER95]. The presence of active attacks means that unauthenticated source routing, either unidirectional (receive-only) or with replies following the original received source route represents a significant security risk unless all received source routed packets are authenticated using the IP Authentication Header or some other cryptologic mechanism. It is noteworthy that the attacks described in [CER95] include a subset of those described in [Bel89]. The use of IP tunneling with AH creates multiple pairs of endpoints that might perform AH processing. Implementers and administrators should carefully consider the impacts of tunneling on authenticity of the received tunneled packets. ACKNOWLEDGEMENTS This document benefited greatly from work done by Bill Simpson, Perry Metzger, and Phil Karn to make general the approach originally defined by the author for SIP, SIPP, and finally IPv6. The basic concept here is derived in large part from the SNMPv2 Security Protocol work described in [GM93]. Steve Bellovin, Steve Deering, Frank Kastenholz, Dave Mihelcic, and Hilarie Orman provided thoughtful critiques of early versions of this note. Francis Dupont discovered and pointed out the security issue with ICMP in low IP MTU links that is noted just above. REFERENCES [Atk95a] Atkinson, R., "Security Architecture for the Internet Protocol", RFC 1825, NRL, August 1995. [Atk95b] Atkinson, R., "IP Encapsulating Security Payload", RFC 1827, NRL, August 1995. [Bel89] Steven M. Bellovin, "Security Problems in the TCP/IP Protocol Suite", ACM Computer Communications Review, Vol. 19, No. 2, March 1989. Atkinson Standards Track [Page 11] RFC 1826 IP Authentication Header August 1995 [BCCH94] Braden, R., Clark, D., Crocker, S., and C. Huitema, "Report of IAB Workshop on Security in the Internet Architecture", RFC 1636, USC/Information Sciences Institute, MIT, Trusted Information Systems, INRIA, June 1994, pp. 21-34. [CER95] Computer Emergency Response Team (CERT), "IP Spoofing Attacks and Hijacked Terminal Connections", CA-95:01, January 1995. Available via anonymous ftp from info.cert.org in /pub/cert_advisories. [GM93] Galvin J., and K. McCloghrie, "Security Protocols for version 2 of the Simple Network Management Protocol (SNMPv2)", RFC 1446, Trusted Information Systems, Hughes LAN Systems, April 1993. [Hin94] Bob Hinden (Editor), Internet Protocol version 6 (IPv6) Specification, Work in Progress, October 1994. [Ken91] Kent, S., "US DoD Security Options for the Internet Protocol", RFC 1108, BBN Communications, November 1991. [Kno93] Knowles, Stev, "IESG Advice from Experience with Path MTU Discovery", RFC 1435, FTP Software, March 1993. [MS95] Metzger, P., and W. Simpson, "IP Authentication with Keyed MD5", RFC 1828, Piermont, Daydreamer, August 1995. [MD90] Mogul, J., and S. Deering, "Path MTU Discovery", RFC 1191, DECWRL, Stanford University, November 1990. [STD-2] Reynolds, J., and J. Postel, "Assigned Numbers", STD 2, RFC 1700, USC/Information Sciences Institute, October 1994. [Riv92] Rivest, R., "MD5 Digest Algorithm", RFC 1321, MIT and RSA Data Security, Inc., April 1992. [VK83] V.L. Voydock & S.T. Kent, "Security Mechanisms in High-level Networks", ACM Computing Surveys, Vol. 15, No. 2, June 1983. Atkinson Standards Track [Page 12] RFC 1826 IP Authentication Header August 1995 DISCLAIMER The views and specification here are those of the author and are not necessarily those of his employer. The Naval Research Laboratory has not passed judgement on the merits, if any, of this work. The author and his employer specifically disclaim responsibility for any problems arising from correct or incorrect implementation or use of this specification. AUTHOR INFORMATION Randall Atkinson Information Technology Division Naval Research Laboratory Washington, DC 20375-5320 USA Phone: (202) 767-2389 Fax: (202) 404-8590 EMail: atkinson@itd.nrl.navy.mil Atkinson Standards Track [Page 13]