A Quick Crash Detection Method for IKE
draft-nir-ike-qcd-01.txt

Status of this Memo

By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts.

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.”

The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt.

The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html.

This Internet-Draft will expire on January 15, 2009.

Abstract

This document describes an extension to the IKEv2 protocol that allows for faster crash recovery using a saved token.

When an IPsec tunnel between two IKEv2 implementations is disconnected due to a restart of one peer, it can take as much as several minutes for the other peer to discover that the reboot has occurred, thus delaying recovery. In this text we propose an extension to the protocol, that allows for recovery immediately following the reboot.

Table of Contents

1.  Introduction
    1.1.  Conventions Used in This Document
2.  RFC 4306 Crash Recovery
3.  Protocol Outline
4.  Stateless Variant Outline
    4.1.  Introducing CHECK_SPI
    4.2.  Stateless Recovery
    4.3.  Wait before rekey
    4.4.  Throttling and Dampening
        4.4.1.  Invalid SPI throttling
        4.4.2.  Dampening
        4.4.3.  User controls
5.  Formats and Exchanges
    5.1.  Notification Format
    5.2.  check_fmt
    5.3.  Stateless IKE Recovery VendorID
    5.4.  Authentication Exchange
    5.5.  Informational Exchange
6.  Token Generation and Verification
    6.1.  A Stateless Method of Token Generation
    6.2.  Token Lifetime
7.  Backup Gateways
8.  Alternative Solutions
    8.1.  Initiating a new IKE SA
    8.2.  Birth Certificates
9.  Interaction with IFARE
10.  Operational Considerations
    10.1.  Who should implement this specification
    10.2.  Response to unknown child SPI
    10.3.  Stateless IKE Recovery cookie
11.  Security Considerations
    11.1.  Security Considerations for the Stateful Method
    11.2.  Security Considerations for the Stateless Method
12.  IANA Considerations
13.  Acknowledgements
14.  Change Log
    14.1.  Changes from draft-nir-ike-qcd-00
    14.2.  Changes from draft-nir-qcr-00
15.  References
    15.1.  Normative References
    15.2.  Informative References
§  Authors' Addresses
§  Intellectual Property and Copyright Statements

1.  Introduction

IKEv2, as described in [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.) has a method for recovering from a reboot of one peer. As long as traffic flows in both directions, the rebooted peer should re-establish the tunnels immediately. However, in many cases the rebooted peer is a VPN gateway that protects only servers, or else the non-rebooted peer has a dynamic IP address. In such cases, the rebooted peer will not be able to re-establish the tunnels. Section 2 (RFC 4306 Crash Recovery) describes how recovery works under RFC 4306, and explains why it takes several minutes.

The method proposed here, is to send a token in the IKE_AUTH exchange that establishes the tunnel. That token can be stored on the peer as part of the IKE SA. After a reboot, the rebooted implementation can re-generate the token, and send it to the non-rebooted peer so as to delete the IKE SA. Deleting the IKE SA results is a quick re-establishment of the IPsec tunnels. This is described in Section 3 (Protocol Outline).

Finally, Section 4 (Stateless Variant Outline) describes a variant that does not require storing state on the non-rebooted peer, but does require an extra round-trip.

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] (Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels,” March 1997.).

The term "token" refers to an octet string that an implementation can generate using only the IKE SPIs as input. A conforming implementation MUST be able to generate the same token from the same input even after rebooting.

The term "token maker" refers to an implementation that generates a token and sends it to the peer in the IKE_AUTH exchange.

The term "token taker" refers to an implementation that stores such a token or a digest thereof, after receiving it in an IKE_AUTH exchange.

2.  RFC 4306 Crash Recovery

When one peer reboots, the other peer does not get any notification, so IPsec traffic can still flow. The rebooted peer will not be able to decrypt it, however, and the only remedy is to send an unprotected INVALID_SPI notification as described in section 3.10.1 of [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.). That section also describes the processing of such a notification: "If this Informational Message is sent outside the context of an IKE_SA, it should be used by the recipient only as a "hint" that something might be wrong (because it could easily be forged)."

Since the INVALID_SPI can only be used as a hint, the non-rebooted peer has to determine whether the IPsec SA, and indeed the parent IKE SA are still valid. The method of doing this is described in section 2.4 of [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.). This method, called "liveness check" involves sending a protected empty INFORMATIONAL message, and awaiting a response. This procedure is sometimes referred to as "Dead Peer Detection" or DPD.

Section 2.4 does not mandate how many times the liveness check message should be retransmitted, or for how long, but does recommend the following: "It is suggested that messages be retransmitted at least a dozen times over a period of at least several minutes before giving up on an SA". Clearly, implementations differ, but all will take a significant amount of time.

3.  Protocol Outline

Supporting implementations will send a notification, called a "QCD token", as described in Section 5.1 (Notification Format) in the last packets of the IKE_AUTH exchange. These are the final request and final response that contain the AUTH payloads. The generation of these tokens is a local matter for implementations, but considerations are described in Section 6 (Token Generation and Verification). Implementations that send such a token will be called "token makers".

A supporting implementation receiving such a token SHOULD store it as part of the IKE SA. Implementations that support this part of the protocol will be called "token takers". Section 10.1 (Who should implement this specification) has considerations for which implementations need to be token takers, and which should be token makers. Implementation that are not token takers will silently ignore QCD tokens.

When a token maker receives a protected IKE request message with unknown IKE SPIs, it MUST generate a new token that is identical to the previous token, and send it to the requesting peer in an unprotected IKE message as described in Section 5.5 (Informational Exchange).

When a token taker receives the QCD token in an unprotected notification, it MUST verify that the TOKEN_SECRET_DATA matches the token stored in the matching the IKE SA. If the verification fails, or if the IKE SPIs in the message do not match any existing IKE SA, it SHOULD log the event. If it succeeds, it MUST delete the IKE SA associated with the IKE_SPI fields, and all dependant child SAs. This event MAY also be logged. The token taker MUST accept such tokens from any address, so as to allow different kinds of high-availability configuration of the token maker.

A supporting token taker MAY immediately create new SAs using an Initial exchange, or it may wait for subsequent traffic to trigger the creation of new SAs.

There is ongoing work on IKEv2 Session Resumption [resumption] (Sheffer, Y., Tschofenig, H., Dondeti, L., and V. Narayanan, “IPsec Gateway Failover Protocol,” March 2008.). See Section 9 (Interaction with IFARE) for a short discussion about this protocol's interaction with session resumption.

4.  Stateless Variant Outline

Sometimes, a QCD token is not available to the non-rebooted implementation. This can happen for several reasons:

• Perhaps the rebooted peer has not implemented the "token maker" part of the protocol.
• Perhaps the non-rebooted peer is resource-constrained, and cannot spare the memory needed to save the token, so it did not implement the "token taker" part of the protocol.

In such cases, we also define a stateless variant of the protocol, that does not require any state on the non-rebooted peer, but does require an extra round-trip.

A supporting implementation will advertise this capability with a special VID payload as defined in Section 5.3 (Stateless IKE Recovery VendorID). When such an implementation reboots and sends an INVALID_SPI or INVALID_IKE_SPI notification to the non-rebooted peer, which has no QCD token, the non-rebooted peer uses a CHECK_SPI notification (see Section 4.1 (Introducing CHECK_SPI)) to poll its peer about whether or not the SPI is actually invalid.

4.1.  Introducing CHECK_SPI

In order to achieve stateless IKE recovery, this memo introduces a new notify type called CHECK_SPI. The CHECK_SPI payload carries an SPI (IKE_SA or Child SA) and one of three sub-types (QUERY, ACK, NACK). The semantic of the CHECK_SPI subtypes is the following:

• QUERY: a peer queries the remote peer SA DB for the presence of the SA whose value is in the payload.
• ACK: a peer confirms it has the SA specified in the payload.
• NACK: a peer confirms it does not have the SA specified in the payload.

The payload format of the CHECK_SPI notify is covered in Section 5.2 (check_fmt).

4.2.  Stateless Recovery

After receiving the INVALID_SPI or INVALID_IKE_SPI notifications, the non-rebooted peer (called Peer Y in the figure) will send an unprotected IKE message as follows. Note that Peer Y MUST NOT send this unless Peer X has advertised this capability in the IKE_AUTH exchange.

   Peer X                                                  Peer Y

             HDR(A,B) INVALID_IKE_SPI(A,B)
            -------------------------------------------->

             HDR(A,B) CHECK_SPI(QUERY,(A,B)), N(Cookie)
            <--------------------------------------------

             HDR(A,B) CHECK_SPI(ACK|NACK,(A,B)), N(Cookie)
            -------------------------------------------->

In this figure, A & B represent the IKE SPIs, and the Cookie is a stateless cookie with similar considerations as the stateless cookie described in section 2.6 of RFC 4306. The cookie SHOULD depend on the IKE SPIs and a saved secret.

A similar exchange happens when the peer sends an INVALID_SPI notification:

   Peer X                                                  Peer Y

             HDR(0,0) INVALID_SPI(a)
            -------------------------------------------->

             HDR(A,B) CHECK_SPI(QUERY,(A,B)), N(Cookie)
            <--------------------------------------------

             HDR(A,B) CHECK_SPI(ACK|NACK,(A,B)), N(Cookie)
            -------------------------------------------->

The difference here is that Peer Y had to locate the IKE SPIs associated with the SPI mentioned in the INVALID_SPI notification.

4.3.  Wait before rekey

There exists a particular attack where a man-in-the-middle can snoop and inject traffic but can not block or drop packets. This attack can spoof INVALID_SPI (allegedly from X), forcing a CHECK_SPI(QUERY) from Y. The attacker would spoof back CHECK_SPI(NACK) to force an undue rekey. Since the attacker can not block packets, the INVALID_SPI will also reach Alice, who will reply with CHECK_SPI(ACK).

Y receives CHECK_SPI(NACK) first and MAY wait for a few msec before creating a new SA. Y will eventually receive BOTH a CHECK_SPI(ACK) and a CHECK_SPI(NACK), Which is dubious. The SIR process should then stop and log an error, saving the SA.

The process is illustrated below:

      X                 Attacker                Y
                            Inv SPI
                            ------------------>

                               CHECK_SPI(QUERY)
         <-------------------------------------

                            CHECK_SPI(NACK)
                            ------------------> Should rekey
                                                but wait a few msec

         CHECK_SPI(ACK)
         -------------------------------------> Hint of attack
                                                => no rekey

Ideally, the round-trip-time should be measured during the IKE exchange and Y wait for a full RTT before initiating a rekey.

Given that IKE itself is subject to DH computation by a man-in-the-middle, also considering that SA's are dampened after creation (see Section 4.4.2 (Dampening)), the staging complexity and limited interest of this attack makes it rather impractical. An implementation MAY decided to implement this final safety wait but this is strictly optional.

4.4.  Throttling and Dampening

An important aspect of the security in stateless IKE recovery has to do with limiting the CPU utilization. In order to thwart flood types denial of service attacks, strict rate limiting and throttling mechanisms have to be enforced.

All the notifications that are exchanged during IKE recovery SHOULD be rate limited. This paragraph provides information on the way rate limiting should take place.

4.4.1.  Invalid SPI throttling

The sending of all Invalid SPI notifies MUST be rate limited one way or an other. The rate limiting SHOULD be performed on a per peer basis but dynamic state creation SHOULD be avoided as much as possible. A recommended tradeoff is to limit the number of flows that can undergo recovery at one point in time and avoid sending Invalid SPI notifies for flows that are potentially already under recovery.

Invalid SPI rate limiting protects against natural dangling SA occurences. I.e. normal traffic conditions may cause unrecognized SPI's to be received and this message is the most important to protect. Indeed, it is not realistic to send one notification per bad ESP packet received. On high speed links, this could mean thousands of IKE notifies sent for the same offending SPI.

The receiving of unauthenticated Invalid SPI notifies MUST as well be rate limited. Again, the rate limiting SHOULD be performed on a per peer basis without dynamic state creation. In normal circumstances, the peer receiving Invalid SPI notifies has an SA with the peer sendig those notifies and already maintains peer-related data structures that can help in maintaining adequate counters.

Authenticated Invalid SPI notifies can be accepted without throttling.

4.4.2.  Dampening

After one of the following conditions:

• the natural creation or rekey of one or more SA's
• the recovery of one or more SA's
• the failure in recovering an SA owned by the local security gateway
• the logging of an error or warning message involving an SA owned by the local security gateway

The peer with which SA's were created, attempted or against which a log was emitted SHOULD be dampened, which means that all the unauthenticated Invalid SPI and Check SPI messages emitted by that peer MUST be ignored for a chosen duration.

This protection prevents a man-in-the-middle from forcing the fast recreation of SA's and potentially depleting the entropy of systems under attack. It also deals efficently with race conditions that may occur after a rekey.

4.4.3.  User controls

Because throttling at large is related to speed, the network implementation around the security gateways has a major influence on the pertinence of the paremeters controlling rate limiting. It is difficult to provide good absolute values for the rate limiters, considering that these are implementation dependent.

As such, for the sake of fitness in practical deployments, a system implementing this memo MUST provide administrative controls over the rate limiter parameters.

5.  Formats and Exchanges

5.1.  Notification Format

The notification payload called "QCD token" is formatted as follows:

                         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    ! QCD Token Notify Message Type !
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    !                                                               !
    ~                       TOKEN_SECRET_DATA                       ~
    !                                                               !
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

• Protocol ID (1 octet) MUST contain 1, as this message is related to an IKE SA.
• SPI Size (1 octet) MUST be zero, in conformance with [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.).
• QCD Token Notify Message Type (2 octets) - MUST be xxxxx, the value assigned for QCD token notifications. TBA by IANA.
• TOKEN_SECRET_DATA (16-256 octets) contains a generated token as described in Section 6 (Token Generation and Verification).

5.2.  check_fmt

The notification payload called "CHECK_SPI" is formatted as follows:

                         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    ! CHECK_SPI Notify Message Type !
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    ! Operation     !
    +-+-+-+-+-+-+-+-+

• Protocol ID (1 octet) MUST contain 1, as this message is related to an IKE SA.
• SPI Size (1 octet) MUST be zero, in conformance with [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.).
• CHECK_SPI Notify Message Type (2 octets) - MUST be xxxxx, the value assigned for CHECK_SPI notifications. TBA by IANA.
• Operation (1 Octet) - This field determines the operation being performed (Query, Reply_ACK, Reply_NACK)

The list of operations and their corresponding value:

• Query: 0
• Reply_ACK: 1
• NACK: 2

5.3.  Stateless IKE Recovery VendorID

The stateless IKE recovery VendorID or SIR_VID is as follows:

"SIR STATELESS" hex: 53 49 52 20 53 54 41 54 45 4c 45 53 53

This VendorID payload MUST be sent in the first IKE_AUTH message of any implementation that supports the stateless variant of this protocol.

5.4.  Authentication Exchange

For clarity, only the EAP version of an AUTH exchange will be presented here. The non-EAP version is very similar. The figure below is based on appendix A.3 of [RFC4718] (Eronen, P. and P. Hoffman, “IKEv2 Clarifications and Implementation Guidelines,” October 2006.).

 first request       --> IDi,
                         [N(INITIAL_CONTACT)],
                         [[N(HTTP_CERT_LOOKUP_SUPPORTED)], CERTREQ+],
                         [IDr],
                         [CP(CFG_REQUEST)],
                         [N(IPCOMP_SUPPORTED)+],
                         [N(USE_TRANSPORT_MODE)],
                         [N(ESP_TFC_PADDING_NOT_SUPPORTED)],
                         [N(NON_FIRST_FRAGMENTS_ALSO)],
                         SA, TSi, TSr,
                         [V(SIR_VID)]
                         [V+]

 first response      <-- IDr, [CERT+], AUTH,
                         EAP,
                         [V(SIR_VID)]
                         [V+]

                   / --> EAP
 repeat 1..N times |
                   \ <-- EAP

 last request        --> AUTH
                         [N(QCD_TOKEN)]

 last response       <-- AUTH,
                         [N(QCD_TOKEN)]
                         [CP(CFG_REPLY)],
                         [N(IPCOMP_SUPPORTED)],
                         [N(USE_TRANSPORT_MODE)],
                         [N(ESP_TFC_PADDING_NOT_SUPPORTED)],
                         [N(NON_FIRST_FRAGMENTS_ALSO)],
                         SA, TSi, TSr,
                         [N(ADDITIONAL_TS_POSSIBLE)],
                         [V+]

Note that the QCD_TOKEN notification is marked as optional because it is not required by this specification that every implementation be both token maker and token taker. If only one peer sends the QCD token, then a reboot of the other peer will not be recoverable by this method. This may be acceptable if traffic typically originates from the other peer.

In any case, the lack of a QCD_TOKEN notification MUST NOT be taken as an indication that the peer does not support this standard. Conversely, if a peer does not understand this notification, it will simply ignore it. Therefore a peer MAY send this notification freely, even if it does not know whether the other side supports it.

5.5.  Informational Exchange

This QCD_TOKEN notification is unprotected, and is sent as a response to a protected IKE request, which uses an IKE SA that is unknown.

         request             --> N(INVALID_IKE_SPI), N(QCD_TOKEN)+

         response            <--

If child SPIs are persistently mapped to IKE SPIs as described in Section 10.2 (Response to unknown child SPI), we may get the following exchange in response to an ESP or AH packet.

         request             --> N(INVALID_SPI), N(QCD_TOKEN)+

         response            <--

The QCD_TOKEN and INVALID_IKE_SPI notifications are sent together to support both implementations that conform to this specification and implementations that don't. Similar to the description in section 2.21 of [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.), The IKE SPI and message ID fields in the packet headers are taken from the protected IKE request.

To support a periodic rollover of token generation constants, the token taker MUST support at least four QCD_TOKEN notifications in a single packet. The token is considered verified if any of the QCD_TOKEN notifications matches. The token maker MAY generate up to four QCD_TOKEN notifications, based on several generations of keys.

If the QCD_TOKEN verifies OK, an empty response MUST be sent. If the QCD_TOKEN cannot be validated, a response SHOULD NOT be sent. Section 6 (Token Generation and Verification) defines token verification.

6.  Token Generation and Verification

No token generation method is mandated by this document. A method is documented in Section 6.1 (A Stateless Method of Token Generation), but only serves as an example.

The following lists the requirements from a token generation mechanism:

• Tokens MUST be at least 16 octets log, and no more than 128 octets long, to facilitate storage and transmission. Tokens SHOULD be indistinguishable from random data.
• It should not be possible for an external attacker to guess the QCD token generated by an implementation. Cryptographic mechanisms such as PRNG and hash functions are RECOMMENDED.
• The token maker, MUST be able to re-generate or retrieve the token based on the IKE SPIs even after it reboots.

6.1.  A Stateless Method of Token Generation

This describes a stateless method of generating a token:

• At installation or immediately after the first boot of the IKE implementation, 32 random octets are generated using a secure random number generator or a PRNG.
• Those 32 bytes, called the "QCD_SECRET", are stored in non-volatile storage on the machine, and kept indefinitely.
• The TOKEN_SECRET_DATA is calculated as follows:

  
           TOKEN_SECRET_DATA = HASH(QCD_SECRET | SPI-I | SPI-R)
  
  

• If key rollover is required by policy, the implementation MAY periodically generate a new QCD_SECRET and keep up to 3 previous generations. When sending an unprotected QCD_TOKEN, as many as 4 notification payloads may be sent, each from a different QCD_SECRET.

6.2.  Token Lifetime

The token is associated with a single IKE SA, and SHOULD be deleted by the token taker when the SA is deleted or expires. More formally, the token is associated with the pair (SPI-I, SPI-R).

7.  Backup Gateways

Making crash recovery quick is important, but since rebooting a gateway takes a non-zero amount of time, many implementations choose to have a stand-by gateway ready to take over as soon as the primary gateway fails for any reason.

If such a configuration is available, it is RECOMMENDED that the stand-by gateway be able to generate the same token as the active gateway. if the method described in Section 6.1 (A Stateless Method of Token Generation) is used, this means that the QCD_SECRET field is identical in both gateways. This has the effect of having the crash recovery available immediately.

8.  Alternative Solutions

8.1.  Initiating a new IKE SA

Instead of sending a QCD token, we could have the rebooted implementation start an Initial exchange with the peer, including the INITIAL_CONTACT notification. This would have the same effect, instructing the peer to erase the old IKE SA, as well as establishing a new IKE SA with fewer rounds.

The disadvantage here, is that in IKEv2 an authentication exchange MUST have a piggy-backed Child SA set up. Since our use case is such that the rebooted implementation does not have traffic flowing to the peer, there are no good selectors for such a Child SA.

Additionally, when authentication is asymmetric, such as when EAP is used, it is not possible for the rebooted implementation to initiate IKE.

8.2.  Birth Certificates

Here we should explain why not Birth Certificates.

9.  Interaction with IFARE

IFARE, specified in [resumption] (Sheffer, Y., Tschofenig, H., Dondeti, L., and V. Narayanan, “IPsec Gateway Failover Protocol,” March 2008.) proposes to make setting up a new IKE SA consume less computing resources. This is particularly useful in the case of a remote access gateway that has many tunnels. A failure of such a gateway would require all these many remote access clients to establish an IKE SA either with the rebooted gateway or with a backup gateway. This tunnel re-establishment should occur within a short period of time, creating a burden on the remote access gateway. IFARE addresses this problem by having the clients store an encrypted derivative of the IKE SA for quick re-establishment.

What IFARE does not help, is the problem of detecting that the peer gateway has failed. A failed gateway may go undetected for as long as the lifetime of a child SA, because IPsec does not have packet acknowledgement. Before establishing a new IKE SA using IFARE, a client MUST ascertain that the gateway has indeed failed. This could be done using either a liveness check (as in RFC 4306) or using the QCD tokens described in this document.

A remote access client conforming to both specifications will store QCD tokens, as well as the IFARE state, if provided by the gateway. A remote access gateway conforming to both specifications will generate a QCD token for the client. When the gateway reboots, the client will discover this in either of two ways:

1. The client does regular liveness checks, or else the time for some other IKE exchange has come. Since the gateway is still down, the IKE times out after several minutes. In this case QCD does not help.
2. Either the primary gateway or a backup gateway (see Section 7 (Backup Gateways)) is ready and sends a QCD token to the client. In that case the client will quickly re-establish the IPsec tunnel, either with the rebooted primary gateway, the backup gateway as described in this document or another gateway as described in [resumption] (Sheffer, Y., Tschofenig, H., Dondeti, L., and V. Narayanan, “IPsec Gateway Failover Protocol,” March 2008.)

The full combined protocol looks like this:

     Initiator                Responder
     -----------              -----------
    HDR, SAi1, KEi, Ni  -->

                        <--    HDR, SAr1, KEr, Nr, [CERTREQ]

    HDR, SK {IDi, [CERT,]
    [CERTREQ,] [IDr,]
    AUTH, N(QCD_TOKEN)
    SAi2, TSi, TSr,
    N(TICKET_REQUEST)}  -->
                        <--    HDR, SK {IDr, [CERT,] AUTH, SAr2, TSi,
                               TSr, N(TICKET_OPAQUE)
                               [,N(TICKET_GATEWAY_LIST)]}

             ---- Reboot -----

    HDR, {}             -->
                        <--  HDR, N(QCD_Token)

    HDR, Ni, N(TICKET_OPAQUE),
    [N+,], SK {IDi, [IDr,]
    SAi2, TSi, TSr,
    [CP(CFG_REQUEST)]}  -->
                        <--  HDR, SK {IDr, Nr, SAr2, [TSi, TSr],
                             [CP(CFG_REPLY)]}

10.  Operational Considerations

10.1.  Who should implement this specification

Throughout this document, we have referred to reboot time alternatingly as the time that the implementation crashes and the time when it is ready to process IPsec packets and IKE exchanges. Depending on the hardware and software platforms and the cause of the reboot, rebooting may take anywhere from a few seconds to several minutes. If the implementation is down for a long time, the benefit of this protocol extension are reduced. For this reason critical systems should implement backup gateways as described in Section 7 (Backup Gateways). Note that the lower-case "should" in the previous sentence is intentional, as we do not specify this in the sense of RFC 2119.

Implementing the "token maker" side of QCD makes sense for IKE implementation where protected connections originate from the peer, such as inter-domain VPNs and remote access gateways. Implementing the "token taker" side of QCD makes sense for IKE implementations where protected connections originate, such as inter-domain VPNs and remote access clients.

To clarify the requirements:

• A remote-access client MUST be a token taker and MAY be a token maker.
• A remote-access gateway MAY be a token taker and MUST be a token maker.
• An inter-domain VPN gateway MUST be both token maker and token taker.

In order to limit the effects of DoS attacks, a token taker SHOULD limit the rate of QCD_TOKENs verified from a particular source.

If excessive amounts of IKE requests protected with unknown IKE SPIs arrive at a token maker, the IKE module SHOULD revert to the behavior described in section 2.21 of [RFC4306] (Kaufman, C., “Internet Key Exchange (IKEv2) Protocol,” December 2005.) and either send an INVALID_IKE_SPI notification, or ignore it entirely.

10.2.  Response to unknown child SPI

After a reboot, it is more likely that an implementation receives IPsec packets than IKE packets. In that case, the rebooted implementation will send an INVALID_SPI notification, triggering a liveness check. The token will only be sent in a response to the liveness check, thus requiring an extra round-trip.

To avoid this, an implementation that has access to non-volatile storage MAY store a mapping of child SPIs to owning IKE SPIs. If such a mapping is available and persistent across reboots, the rebooted implementation MAY respond to the IPsec packet with an INVALID_SPI notification, along with the appropriate QCD_Token notifications. A token taker SHOULD verify the QCD token that arrives with an INVALID_SPI notification the same as if it arrived with the IKE SPIs of the parent IKE SA.

However, a persistent storage module might not be updated in a timely manner, and could be populated with IKE SPIs that have already been rekeyed. A token taker MUST NOT take an invalid QCD Token sent along with an INVALID_SPI notification as evidence that the peer is either malfunctioning or attacking, but it SHOULD limit the rate at which such notifications are processed.

10.3.  Stateless IKE Recovery cookie

The cookie information is chosen by the peer that emits it. As such, the cookie has strictly no meaning for the remote peer and can thus be chosen as seen fit. This section provides recommendations on how to generate and validate those cookies.

When an IKE endpoint X sends an unauthenticated CHECK_SPI, the cookie payload following the notify is computed as follow:

               Cookie = VersionIDofSecret
                        | H( SECRET | CHECK_SPI(..., Query)
                        | ip.src | ip.dst
                        | udp.src | udp.dst)

where

• SECRET is a randomly generated secret known only to the implementation and periodically changed.
• VersionIDofSecret should be changed whenever SECRET is regenerated.
• CHECK_SPI(..., Query) is the content of the CHECK_SPI notify payload where the operation subtype has been set to Query (cf. Section 4.1 (Introducing CHECK_SPI))
• ip.src is the source ip address of the IKE packet.
• ip.dst is the destination ip address of the IKE packet.
• udp.src is the source udp post of the IKE packet.
• udp.dst is the destination udp port of the IKE packet.

Upon reception of a CHECK_SPI notify (ACK or NACK) followed by a N(Cookie), a peer can verify whether this is the reply to a Query it placed by recomputing the cookie and comparing it to the COOKIE in the IKE message.

In order to minimize the range of cryptographic attacks on SECRET, its value SHOULD have a limited life time.

11.  Security Considerations

11.1.  Security Considerations for the Stateful Method

Tokens MUST be hard to guess. This is critical, because if an attacker can guess the token associated with the IKE SA, she can tear down the IKE SA and associated tunnels at will. When the token is delivered in the IKE_AUTH exchange, it is encrypted. When it is sent again in an unprotected notification, it is not, but that is the last time this token is ever used.

An aggregation of some tokens generated by one peer together with the related IKE SPIs MUST NOT give an attacker the ability to guess other tokens. Specifically, if one peer does not properly secure the QCD tokens and an attacker gains access to them, this attacker MUST NOT be able to guess other tokens generated by the same peer. This is the reason that the QCD_SECRET in Section 6.1 (A Stateless Method of Token Generation) needs to be sufficiently long.

The QCD_SECRET MUST be protected from access by other parties. Anyone gaining access to this value will be able to delete all the IKE SAs for this token maker.

The QCD token is sent by the rebooted peer in an unprotected message. A message like that is subject to modification, deletion and replay by an attacker. However, these attacks will not compromise the security of either side. Modification is meaningless because a modified token is simply an invalid token. Deletion will only cause the protocol not to work, resulting in a delay in tunnel re-establishment as described in Section 2 (RFC 4306 Crash Recovery). Replay is also meaningless, because the IKE SA has been deleted after the first transmission.

11.2.  Security Considerations for the Stateless Method

IKE recovery self-protection is discussed all along the document and contains many mechanism to thwart denial of service attacks.

IKE recovery is subject to a man-in-the-middle attack that can let the attacker trigger a renegotiation. It has to be noticed that an attacker able to block ESP and/or IKE packets can cause IKE itself to also tear down and trigger a rekey of IKE SA's. With throttling and dampening enabled, IKE recovery is able to reduce the amount of rekeys/negotiations to as low a rate as IKEv2.

Overall, IKE Recovery is not more vulnerable than IKEv2 and even improves on the security of IKEv2 by resynchronizing SA's more rapidly which is important with dynamic polices.

12.  IANA Considerations

IANA is requested to assign a notify message type from the error types range (43-8191) of the "IKEv2 Notify Message Types" registry with name "QUICK_CRASH_DETECTION".

IANA is requested to assign a notify message type from the status types range (16406-40959) of the "IKEv2 Notify Message Types" registry with name "CHECK_SPI".

13.  Acknowledgements

We would like to thank Hannes Tschofenig and Yaron Sheffer for their comments about IFARE.

14.  Change Log

This section lists all changes in this document

NOTE TO RFC EDITOR : Please remove this section in the final RFC

14.1.  Changes from draft-nir-ike-qcd-00

• Merged proposal with draft-detienne-ikev2-recovery [recovery] (Detienne, F. and P. Sethi, “Safe IKE Recovery,” June 2008.)
• Changed the protocol so that the rebooted peer generates the token. This has the effect, that the need for persistent storage is eliminated.
• Added discussion of birth certificates.

14.2.  Changes from draft-nir-qcr-00

• Changed name to reflect that this relates to IKE. Also changed from quick crash recovery to quick crash detection to avoid confusion with IFARE.
• Added more operational considerations.
• Added interaction with IFARE.
• Added discussion of backup gateways.

15.  References

15.1. Normative References

15.2. Informative References

Authors' Addresses

Full Copyright Statement

Copyright © The IETF Trust (2008).

This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights.

This document and the information contained herein are provided on an “AS IS” basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79.

Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr.

The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org.