RFC 6867

An Internet Key Exchange Protocol Version 2 (IKEv2) Extension to Support EAP Re-authentication Protocol (ERP)

📅 published January 2013, from ietf

RFC 6867 ✨ An IKEv2 Extension to Support EAP Re-authentication Protocol (ERP) 🎀

The vibe check: Y. Nir (Check Point) and Q. Wu (Huawei) wrote this one. It's an Experimental doc, ISSN 2070-1721, dated January 2013. It updates RFC 5996 and doesn't obsolete anything. Request for Comments: 6867, from the Internet Engineering Task Force (IETF). 💅

Abstract

okay so this doc updates IKEv2 from RFC 5996 🦋 It lets an IKE Security Association (SA) get created and authenticated using the EAP Re-authentication Protocol extension from RFC 6696. Basically, you can skip the whole long login again and just prove you're you. 🫶

Status of This Memo

ngl, this part is the fine print: this is not an Internet Standards Track spec. It's published so people can look at it, try it out in experimental implementations, and evaluate it. 🧪

This is an Experimental Protocol for the Internet community, a product of the IETF, and it reflects the IETF community's consensus. It got public review and was approved by the Internet Engineering Steering Group (IESG). Not every IESG-approved doc is up for any level of Internet Standard though (see Section 2 of RFC 5741, bestie). 📜

Want the current status, errata, or a way to give feedback? Check http://www.rfc-editor.org/info/rfc6867 ✨

© 2013 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. Please actually read those, because they spell out your rights and restrictions for 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 they come without warranty as described in the Simplified BSD License. ⚖️

1. Introduction ✨

okay so IKEv2 (from [RFC5996]) lets the initiator authenticate with an EAP method (Section 2.16). The catch? EAP adds a ton of round trips to set up the IPsec SA, and it can also need the user to do something. 🙈 So a single remote access client making multiple IPsec tunnels to gateways in the same domain gets super annoying.

The EAP Re-authentication Protocol (ERP) from [RFC6696] is the glow-up here 💖 An EAP peer can authenticate to several authenticators while only doing the full EAP method once. Later logins need fewer round trips and zero user interaction. Iconic. 👑

Put the two together and a remote access IPsec client can make several tunnels to different gateways in one domain. It can also reuse keys from other EAP contexts, like network access within that same domain, to connect to VPN gateways there without any fuss. 🧚

It also means faster setup for new tunnels after old ones get torn down, like after a network outage, device suspension, or a temporary move out of range. Lowkey it's similar to the session resumption mechanism in [RFC5723] 🌸 The one difference: instead of a ticket stored by the client, the re-authentication Master Session Key (rMSK) (see Section 4.6 of [RFC6696]) is the session key, stored on both the client and the Authentication, Authorization, and Accounting (AAA) server. 🔑

1.1. Conventions Used in This Document 🎀

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 [RFC2119]. 📖

2. Usage Scenarios 🍓

this work is motivated by the following scenarios:

  • Multiple tunnels for a single remote access VPN client. Picture a company with offices in New York City, Paris, and Shanghai 🌍 For historical reasons, the email server lives in Paris, most of the intranet servers are in Shanghai, and the finance department servers are in New York City. An employee on a remote access VPN may need servers from all three places. Sure, they could connect to one gateway and have it route requests to the others (maybe through a site to site VPN), but that's not efficient. It's way better for the client to start three separate tunnels. What's not cute is making the user type in a password three times. 🙅‍♀️
  • Roaming. In the era of phones and tablets, people bounce from the office wireless LAN (where access may be granted through 802.1x) to a cellular network, where a VPN is necessary, and back again 📱 Both the VPN server and the 802.1x access point are authenticators connecting to the same AAA servers. So it makes sense to make the switch smooth, without making the user do anything. The device still needs to detect whether it's inside the protected network, in which case it should not use the VPN. That part is beyond the scope of this document though. [SECBEAC] is a now-abandoned attempt at it 🥀
  • Resumption. If a device gets disconnected from an IKE peer, ERP can reconnect it to the same gateway with no user intervention. Main character energy. ✨

3. Protocol Outline 🦋

okay so supporting the EAP Re-authentication Extension (ERX) requires an EAP payload in the first IKE_AUTH request. That's a deviation from the rules in [RFC5996], so support has to be indicated through a Notify payload in the IKE_SA_INIT response. This Notify does the same job as the EAP-Initiate/Re-auth-Start message of ERX, as specified in Section 5.3.1 of [RFC6696]. The "Domain Name" field holds the content of the Domain-Name TLV, as specified in Section 5.3.1.1 of that same document. 🧩

A supporting initiator that has unexpired keys for this domain sends the EAP-Initiate/Re-auth message in an EAP payload in the first IKE_AUTH request. 🚀

The responder sends the EAP payload content to a backend AAA server. If that server has a valid rMSK for the session, it sends those along with an EAP-Finish/Re-auth message. The responder then forwards the EAP-Finish/Re-auth message to the initiator in an EAP payload within the first IKE_AUTH response. 📬

Next, the initiator sends an additional IKE_AUTH request that includes the AUTH payload. That payload is calculated using the rMSK in the role of the MSK, as described in Sections 2.15 and 2.16 of [RFC5996]. The responder replies similarly, and the IKE_AUTH exchange is finished. Done and dusted. 💖

If the backend AAA server doesn't have valid keys for the Re-auth-Start message, it sends back a normal EAP request and no rMSK key. From there, EAP flow continues as in [RFC5996]. 🙃

The figure below is adapted from Appendixes C.1 and C.3 of [RFC5996], with most of the optional payloads removed. Note that the EAP-Initiate/Re-auth message has been added. 🎀

   IKE_SA_INIT Exchange:
   | init request         --> SA, KE, Ni,
   |
   | init response       <-- SA, KE, Nr,
   |                         N[ERX_SUPPORTED]

   IKE_AUTH Exchanges:
   | first request       --> EAP(EAP-Initiate/Re-auth),
   |                         IDi,
   |                         SA, TSi, TSr
   |
   | first response      <-- IDr, [CERT+], AUTH,
   |                         EAP(EAP-Finish/Re-auth)
   |
   | last request        --> AUTH
   |
   | last response       <-- AUTH,
   |                         SA, TSi, TSr

The IDi payload MUST have ID Type ID_RFC822_ADDR, and the data field MUST contain the same value as the KeyName-NAI TLV in the EAP-Initiate/Re-auth message. See Section 3.2 for details. 🫶

3.1. Clarification about EAP Codes 🔍

Section 3.16 of [RFC5996] lists the EAP codes in EAP messages carried in EAP payloads. The list only goes up to 4, and it's not clear whether that list is meant to be exhaustive. 🤷‍♀️

To clear things up: an implementation conforming to this specification MUST accept and transmit EAP messages with at least the codes for Initiate and Finish (5 and 6) from Section 5.3 of [RFC6696], on top of the four codes enumerated in [RFC5996]. This document is intentionally silent about any other EAP codes that aren't enumerated in those documents. Not my problem, not this doc's problem. 💅🌈

3.2. Username in the Protocol ✨

okay so the authors and the HOKEY and IPsecME folks all figured every use case for this IKE extension has ONE backend AAA server doing both the authentication and the re-authentication 🏠. the logic is that IKE runs over the Internet and would naturally connect to the user's home network, so that's the default vibe.

this section is for when that ISN'T the case 👀.

Section 5.3.2 of [RFC6696] describes the EAP-Initiate/Re-auth packet. with IKEv2, it rides in the first IKE_AUTH request 📦. this packet has the KeyName-NAI TLV, and that TLV holds the username used in authentication. it gets relayed to the AAA server in the AccessRequest message and comes back from the AAA server in the AccessAccept message 💌.

the username part of the Network Access Identifier (NAI) inside that TLV is the EMSKName [RFC5295], written in hexadecimal digits 🔢. the domain part is the domain name of the user's home domain 🏡. the username part is ephemeral, meaning a new one gets generated for every full authentication. so it's NOT a great basis for policy decisions, and it's also a poor source of user identification for logging 🙃.

so instead, it's up to the IPsec gateway implementation to make policy decisions based on other factors 💅. this list is by no means exhaustive:

  • sometimes the home domain name is enough. if everyone from a particular home domain gets the same authorization, then policy doesn't depend on the real username 🌸. meaningful logs can still happen by correlating VPN gateway IKE events with AAA servers access records.
  • sometimes users get different authorizations based on the groups they belong to 🧚. the AAA server can pass that info to the VPN gateway using the CLASS attribute [RFC2865] in RADIUS and Diameter [RFC6733]. logging again depends on correlating with AAA servers.
  • AAA servers may support extensions that let them talk to their clients (in our case -- the VPN gateway) to push user info 📤. for example, a certain product integrates a RADIUS server with the Lightweight Directory Access Protocol (LDAP) [RFC4511], so a client could query the server over LDAP and get the real record for that user. others may provide this data through vendor-specific extensions to RADIUS or Diameter.

in any case, authorization is a MAJOR issue in deployments if the backend AAA server doing re-authentication is different from the one that handled the original authentication 😭. it's up to the re-authenticating AAA server to provide the info needed for authorization. a conforming implementation of this protocol MAY reject initiators it can't make policy decisions for, for these reasons 🚫.

4. ERX_SUPPORTED Notification 🎀

the Notify payload is as described in [RFC5996] 📝:

                        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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   ! Next Payload  !C!  RESERVED   !         Payload Length        !
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   !  Protocol ID  !   SPI Size    !    ERX Notify Message Type    !
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   !                            Domain Name                        !
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

here's the breakdown, bestie 🫶:

  • Protocol ID (1 octet) - MUST be zero, as this message is related to an IKE SA 🔒.
  • Security Parameter Index (SPI) Size (1 octet) - MUST be zero, in conformance with Section 3.10 of [RFC5996] ✨.
  • ERX Notify Message Type (2 octets) - MUST be 16427, the value assigned for ERX 🔥.
  • Domain Name (variable) - contains the domain name or realm, as these terms are used in [RFC6696] and encoded as ASCII, as specified in [RFC4282] 🌈.

5. Operational Considerations 💜

this spec changes how IKE peers behave, both initiators AND responders 🔄. backend AAA servers aren't changed by it, but they're required to support [RFC6696]. same goes for the EAP client if it's not built into the IKE initiator (like if the EAP client is an operating system component) 🖥️.

this spec is silent about key storage and key lifetimes on either the EAP client or the EAP server 🤐. those topics are covered in Sections 3, 4, and 5 of [RFC6696]. the key lifetime can be communicated from the AAA server to the EAP client through the Lifetime attribute in the EAP-Finish/Re-auth message ⏳. if the server doesn't have a valid key but the client does, regular EAP gets used (see Section 3). this should not happen if lifetimes are communicated 🙈.

in that situation, the IKEv2 initiator / EAP client MAY alert the user and MAY log the event 🚨. and note that this does NOT necessarily mean an attack! it could simply be a loss of state on the AAA server 🧠.

6. Security Considerations 🛡️

okay this part is SERIOUS bestie, so pay attention 💀.

the protocol extension in this doc extends authentication from one EAP context (which may or may not be part of IKEv2) to an IKEv2 context 🔐. successful completion of the protocol proves to the authenticator, which in our case is a VPN gateway, that the supplicant or VPN client authenticated in some other EAP context ✅.

the protocol gives the authenticator the domain name the supplicant authenticated with, but NOT a specific identity 👻. instead, the gateway gets an EMSKName, which is an ephemeral ID. with this variant of IKEv2, the initiator never sends its real identity on the wire, while the server does. this is different from usual IKEv2 practice, where the initiator reveals its identity first 🙃.

if the domain name is enough to make access control decisions, great, that's all you need 🎉. if not, then the gateway needs to find out either the real name or the authorization info for that particular user 🕵️. this may be done using the AAA protocol or by some other federation protocol, which is out of scope for this specification 🚧.

7. IANA Considerations 📋

IANA assigned a notify message type of 16427 from the "IKEv2 Notify Message Types - Status Types" registry, with the name ERX_SUPPORTED 🎀.

8. Acknowledgements 🫶

the authors want to thank Yaron Sheffer for comments and suggested text that helped shape this document 💖.

thanks also to Juergen Schoenwaelder for his OPS-DIR review comments 🌷.

9. References 📚

9.1. Normative References

[RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
           Requirement Levels", BCP 14, RFC 2119, March 1997.

[RFC4282]  Aboba, B., Beadles, M., Arkko, J., and P. Eronen, "The
           Network Access Identifier", RFC 4282, December 2005.

[RFC5295]  Salowey, J., Dondeti, L., Narayanan, V., and M. Nakhjiri,
           "Specification for the Derivation of Root Keys from an
           Extended Master Session Key (EMSK)", RFC 5295,
           August 2008.

[RFC5996]  Kaufman, C., Hoffman, P., Nir, Y., and P. Eronen,
           "Internet Key Exchange Protocol Version 2 (IKEv2)",
           RFC 5996, September 2010.

[RFC6696]  Cao, Z., He, B., Shi, Y., Wu, Q., and G. Zorn, "EAP
           Extensions for the EAP Re-authentication Protocol (ERP)",
           RFC 6696, July 2012.

9.2. Informative References

[RFC2865]  Rigney, C., Willens, S., Rubens, A., and W. Simpson,
           "Remote Authentication Dial In User Service (RADIUS)",
           RFC 2865, June 2000.

[RFC4511]  Sermersheim, J., "Lightweight Directory Access Protocol
           (LDAP): The Protocol", RFC 4511, June 2006.

[RFC5723]  Sheffer, Y. and H. Tschofenig, "Internet Key Exchange
           Protocol Version 2 (IKEv2) Session Resumption", RFC 5723,
           January 2010.

[RFC6733]  Fajardo, V., Arkko, J., Loughney, J., and G. Zorn,
           "Diameter Base Protocol", RFC 6733, October 2012.

[SECBEAC]  Sheffer, Y. and Y. Nir, "Secure Beacon: Securely Detecting
           a Trusted Network", Work in Progress, June 2009.

Authors' Addresses 💌

Yoav Nir
Check Point Software Technologies Ltd.
5 Hasolelim st.
Tel Aviv  67897
Israel

EMail: ynir@checkpoint.com


Qin Wu
Huawei Technologies Co., Ltd.
101 Software Avenue, Yuhua District
Nanjing, JiangSu  210012
China

EMail: sunseawq@huawei.com
📦 where this page comes from