HIP Working Group                                           G. Camarillo
Internet-Draft                                               P. Nikander
Intended status: Experimental                              J. Hautakorpi
Expires: April 29, 2010                                       A. Keranen
                                                             A. Johnston
                                                        October 26, 2009

    HIP BONE: Host Identity Protocol (HIP) Based Overlay Networking

Status of this Memo

   This Internet-Draft is submitted to IETF in full conformance with the
   provisions of BCP 78 and 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-

   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

   The list of Internet-Draft Shadow Directories can be accessed at

   This Internet-Draft will expire on April 29, 2010.

Copyright Notice

   Copyright (c) 2009 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 in effect on the date of
   publication of this document (http://trustee.ietf.org/license-info).
   Please review these documents carefully, as they describe your rights
   and restrictions with respect to this document.

Camarillo, et al.        Expires April 29, 2010                 [Page 1]

Internet-Draft                  HIP BONE                    October 2009


   This document specifies a framework to build HIP (Host Identity
   Protocol)-based overlay networks.  This framework uses HIP to perform
   connection management.  Other functions, such as data storage and
   retrieval or overlay maintenance, are implemented using protocols
   other than HIP.  These protocols are loosely referred to as peer

Table of Contents

   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  3
   2.  Background on HIP  . . . . . . . . . . . . . . . . . . . . . .  3
     2.1.  ID/locator Split . . . . . . . . . . . . . . . . . . . . .  3
       2.1.1.  Identifier Format  . . . . . . . . . . . . . . . . . .  4
       2.1.2.  HIP Base Exchange  . . . . . . . . . . . . . . . . . .  4
       2.1.3.  Locator Management . . . . . . . . . . . . . . . . . .  5
     2.2.  NAT Traversal  . . . . . . . . . . . . . . . . . . . . . .  5
     2.3.  Security . . . . . . . . . . . . . . . . . . . . . . . . .  6
       2.3.1.  DoS Protection . . . . . . . . . . . . . . . . . . . .  6
       2.3.2.  Identifier Assignment and Authentication . . . . . . .  6
       2.3.3.  Connection Security  . . . . . . . . . . . . . . . . .  7
     2.4.  HIP Deployability and Legacy Applications  . . . . . . . .  7
   3.  The HIP BONE Framework . . . . . . . . . . . . . . . . . . . .  8
     3.1.  Peer ID Assignment and Bootstrap . . . . . . . . . . . . .  9
     3.2.  Connection Establishment . . . . . . . . . . . . . . . . . 10
     3.3.  Lightweight Message Exchanges  . . . . . . . . . . . . . . 11
     3.4.  Participating in Multiple Overlays . . . . . . . . . . . . 11
     3.5.  HIP BONE Instantiation . . . . . . . . . . . . . . . . . . 12
   4.  Advantages of Using HIP BONE . . . . . . . . . . . . . . . . . 13
   5.  Architectural Considerations . . . . . . . . . . . . . . . . . 13
   6.  Security Considerations  . . . . . . . . . . . . . . . . . . . 15
   7.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 16
   8.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . . 16
   9.  References . . . . . . . . . . . . . . . . . . . . . . . . . . 16
     9.1.  Normative References . . . . . . . . . . . . . . . . . . . 16
     9.2.  Informative References . . . . . . . . . . . . . . . . . . 17
   Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 18

Camarillo, et al.        Expires April 29, 2010                 [Page 2]

Internet-Draft                  HIP BONE                    October 2009

1.  Introduction

   The Host Identity Protocol (HIP) [RFC5201] defines a new name space
   between the network and transport layers.  HIP provides upper layers
   with mobility, multihoming, NAT (Network Address Translation)
   traversal, and security functionality.  HIP implements the so called
   identifier/locator (ID/locator) split, which implies that IP
   addresses are only used as locators, not as host identifiers.  This
   split makes HIP a suitable protocol to build overlay networks that
   implement identifier-based overlay routing over IP networks, which in
   turn implement locator-based routing.

   The remainder of this document is organized as follows.  Section 2
   provides background information on HIP.  Section 3 describes the HIP
   BONE (HIP-Based Overlay Networking Environment) framework.  Section 4
   discusses some of the advantages derived from using the HIP BONE
   framework.  Finally, before the customary sections, Section 5
   attempts to put the presented proposal into a larger architectural

2.  Background on HIP

   This section provides background on HIP.  Given the tutorial nature
   of this section, readers that are familiar with what HIP provides and
   how HIP works may want to skip it.  All descriptions contain
   references to the relevant HIP specifications where readers can find
   detailed explanations on the different topics discussed in this

2.1.  ID/locator Split

   In an IP network, IP addresses typically serve two roles: they are
   used as host identifiers and as host locators.  IP addresses are
   locators because a given host's IP address indicates where in the
   network that host is located.  IP networks route based on these
   locators.  Additionally, IP addresses are used to identify remote
   hosts.  The simultaneous use of IP addresses as host identifiers and
   locators makes mobility and multihoming complicated.  For example,
   when a host opens a TCP connection, the host identifies the remote
   end of the connection by the remote IP address (plus port).  If the
   remote host changes its IP address, the TCP connection will not
   survive, since the transport layer identifier of the remote end of
   the connection has changed.

   Mobility solutions such as Mobile IP keep the remote IP address from
   changing so that it can still be used as an identifier.  HIP, on the
   other hand, uses IP addresses as only locators and defines a new

Camarillo, et al.        Expires April 29, 2010                 [Page 3]

Internet-Draft                  HIP BONE                    October 2009

   identifier space.  This approach is referred to as the ID/locator
   split and makes the implementation of mobility and multihoming more
   natural.  In the previous example, the TCP connection would be bound
   to the remote host's identifier, which would not change when the
   remote hosts moves to a new IP address (i.e., to a new locator).  The
   TCP connection is able to survive locator changes because the remote
   host's identifier does not change.

2.1.1.  Identifier Format

   HIP uses 128-bit ORCHIDs (Overlay Routable Cryptographic Hash
   Identifiers) [RFC4843] as identifiers.  ORCHIDs look like IPv6
   addresses but cannot collide with regular IPv6 addresses because
   ORCHID spaces are registered with the IANA.  That is, a portion of
   the IPv6 address space is reserved for ORCHIDs.  The ORCHID
   specification allows creating multiple disjoint identifier spaces.
   Each such space is identified by a separate Context Identifier.  The
   Context Identifier can be either drawn implicitly from the context
   the ORCHID is used in or carried explicitly in a protocol.

   HIP defines a native socket API [I-D.ietf-hip-native-api] that
   applications can use to establish and manage connections.
   Additionally, HIP can also be used through the traditional IPv4 and
   IPv6 TCP/IP socket APIs.  Section 2.4 describes how an application
   using these traditional APIs can make use of HIP.  Figure 1 shows all
   these APIs between the application and the transport layers.

                |               Application               |
                | HIP Native API | Traditional Socket API |
                |             Transport Layer             |

                             Figure 1: HIP API

2.1.2.  HIP Base Exchange

   Before two HIP hosts exchange upper-layer traffic, they perform a
   four-way handshake that is referred to as the HIP base exchange.
   Figure 2 illustrates the HIP base exchange.  The initiator sends an
   I1 packet and receives an R1 packet from the responder.  After that,
   the initiator sends an I2 packet and receives an R2 packet from the

Camarillo, et al.        Expires April 29, 2010                 [Page 4]

Internet-Draft                  HIP BONE                    October 2009

                  Initiator                     Responder

                      |             I1             |
                      | -------------------------->|
                      |             R1             |
                      | <--------------------------|
                      |             I2             |
                      | -------------------------->|
                      |             R2             |
                      | <--------------------------|

                        Figure 2: HIP base exchange

   Of course, the initiator needs the responder's locator (or locators)
   in order to send its I1 packet.  The initiator can obtain locators
   for the responder in multiple ways.  For example, according to the
   current HIP specifications the initiator can get the locators
   directly from the DNS [RFC5205] or indirectly by sending packets
   through a HIP rendezvous server [RFC5204].  However, as an
   architecture HIP is open ended, and allows the locators to be
   obtained by any means (e.g., from packets traversing an overlay
   network or as part of the candidate address collection process in a
   NAT traversal scenario).

2.1.3.  Locator Management

   Once a HIP connection between two hosts has been established with a
   HIP base exchange, the hosts can start exchanging higher-layer
   traffic.  If any of the hosts changes its set of locators, it runs an
   update exchange [RFC5206], which consists of three messages.  If a
   host is multihomed, it simply provides more than one locator in its
   exchanges.  However, if both of the end points move at the same time,
   or through some other reason both lose track of the peers' currently
   active locators, they need to resort to using a rendezvous server or
   getting new peer locators by some other means.

2.2.  NAT Traversal

   HIP's NAT traversal mechanism [I-D.ietf-hip-nat-traversal] is based
   on ICE (Interactive Connectivity Establishment)
   [I-D.ietf-mmusic-ice].  Hosts gather address candidates and, as part
   of the HIP base exchange, hosts perform an ICE offer/answer exchange
   where they exchange their respective address candidates.  Hosts
   perform end-to-end STUN [RFC5389] based connectivity checks in order
   to discover which address candidate pairs yield connectivity.

   Even though, architecturally, HIP lies below the transport layer
   (i.e., HIP packets are carried directly in IP packets), in presence

Camarillo, et al.        Expires April 29, 2010                 [Page 5]

Internet-Draft                  HIP BONE                    October 2009

   of NATs, HIP sometimes needs to be tunneled in a transport protocol
   (i.e., HIP packets are carried by a transport protocol such as UDP).

2.3.  Security

   Security is an essential part of HIP.  The following sections
   describe the security-related functionality provided by HIP.

2.3.1.  DoS Protection

   HIP provides protection against DoS (Denial of Service) attacks by
   having initiators resolve a cryptographic puzzle before the responder
   stores any state.  On receiving an I1 packet, a responder sends a
   pre-generated R1 packet that contains a cryptographic puzzle and
   deletes all the state associated with the processing of this I1
   packet.  The initiator needs to resolve the puzzle in the R1 packet
   in order to generate an I2 packet.  The difficulty of the puzzle can
   be adjusted so that, if a receiver is under a DoS attack, it can
   increase the difficulty of its puzzles.

   On receiving an I2 packet, a receiver checks that the solution in the
   packet corresponds to a puzzle generated by the receiver and that the
   solution is correct.  If it is, the receiver processes the I2 packet.
   Otherwise, it silently discards it.

   In an overlay scenario, there are multiple ways how this mechanism
   can be utilized within the overlay.  One possibility is to cache the
   pre-generated R1 packets within the overlay and let the overlay
   directly respond with R1s to I1s.  In that way the responder is not
   bothered at all until the initiator sends an I2 packet, with the
   puzzle solution.  Furthermore, a more sophisticated overlay could
   verify that an I2 packet has a correctly solved puzzle before
   forwarding the packet to the responder.

2.3.2.  Identifier Assignment and Authentication

   As discussed earlier, HIP uses ORCHIDs [RFC4843] as the main
   representation for identifiers.  Potentially, HIP can use different
   types of ORCHIDs as long as the probability of finding collisions
   (i.e., two nodes with the same ORCHID) is low enough.  One way to
   completely avoid this type of collision is to have a central
   authority generate and assign ORCHIDs to nodes.  To secure the
   binding between ORCHIDs and any higher-layer identifiers, every time
   the central authority assigns an ORCHID to a node, it also generates
   and signs a certificate stating who is the owner of the ORCHID.  The
   owner of the ORCHID then includes the corresponding certificate in
   its R1 (when acting as responder) and I2 packets (when acting
   initiator) to prove that it is actually allowed to use the ORCHID

Camarillo, et al.        Expires April 29, 2010                 [Page 6]

Internet-Draft                  HIP BONE                    October 2009

   and, implicitly, the associated public key.

   Having a central authority works well to completely avoid collisions.
   However, having a central authority is impractical in some scenarios.
   As defined today, HIP systems generally use a self-certifying ORCHID
   type called HIT (Host Identity Tag) that does not require a central
   authority (but still allows one to be used).

   A HIT is the hash of a node's public key.  A node proves that it has
   the right to use a HIT by showing its ability to sign data with its
   associated private key.  This scheme is secure due to the so called
   second-preimage resistance property of hash functions.  That is,
   given a fixed public key K1, finding a different public key K2 such
   that hash(K1) = hash(K2) is computationally very hard.  Optimally, a
   preimage attack on the 100-bit hash function used in ORCHIDs will
   take an order of 2^100 operations to be successful, and can be
   expected to take in the average 2^99 operations.  Given that each
   operation requires the attacker to generate a new key pair, the
   attack is completely impractical (see [RFC4843]).

   HIP nodes using HITs as ORCHIDs do not typically use certificates
   during their base exchanges.  Instead, the use a leap-of-faith
   mechanism, similar to SSH, whereby a node authenticates somehow
   remote nodes the first time they connect it and, then, remembers
   their public keys.  While user-assisted leap-of-faith (such as in
   SSH) can be used to facilitate a human-operated offline path (such as
   a telephone call), automated leap-of-faith can be combined with a
   reputation management system to create an incentive to behave.
   However, such considerations go well beyond the current HIP
   architecture and even beyond this proposal.  For the purposes of the
   present document, we merely want to point out that architecturally
   HIP supports both self-generated opportunistic identifiers and
   administratively assigned ones.

2.3.3.  Connection Security

   Once two nodes complete a base exchange between them, the traffic
   they exchange is encrypted and integrity protected.  The security
   mechanism used to protect the traffic is IPsec ESP [RFC5202].
   However, there is ongoing work to specify how to use different
   protection mechanisms.

2.4.  HIP Deployability and Legacy Applications

   As discussed earlier, HIP defines a native socket API
   [I-D.ietf-hip-native-api] that applications can use to establish and
   manage connections.  New applications can implement this API to get
   full advantage of HIP.  However, in most cases, legacy (i.e., non-HIP

Camarillo, et al.        Expires April 29, 2010                 [Page 7]

Internet-Draft                  HIP BONE                    October 2009

   aware) applications [RFC5338] can use HIP through the traditional
   IPv4 and IPv6 socket APIs.

   The idea is that when a legacy IPv6 application tries and obtains a
   remote host's IP address (e.g., by querying the DNS) the DNS resolver
   passes the remote host's ORCHID (which was also stored in the DNS) to
   the legacy application.  At the same time, the DNS resolver stores
   the remote host's IP address internally at the HIP module.  Since the
   ORCHID looks like an IPv6 address, the legacy application treats it
   as such.  It opens a connection (e.g., TCP) using the traditional
   IPv6 socket API.  The HIP module running in the same host as the
   legacy application intercepts this call somehow (e.g., using an
   interception library or setting up the host's routing tables so that
   the HIP module receives the traffic) and runs HIP (on behalf of the
   legacy application) towards the IP address corresponding to the
   ORCHID.  This mechanism works well in almost all cases.  However,
   applications involving referrals (i.e., passing of IPv6 addresses
   between applications) present issues, to be discussed in Section 3
   below.  Additionally, management applications that care about the
   exact IP address format may not work well with such straightforward

   In order to make HIP work through the traditional IPv4 socket API,
   the HIP module passes an LSI (Local Scope Identifier), instead of a
   regular IPv4 address, to the legacy IPv4 application.  The LSI looks
   like an IPv4 address, but is locally bound to an ORCHID.  That is,
   when the legacy application uses the LSI in a socket call, the HIP
   module intercepts it and replaces the LSI with its corresponding
   ORCHID.  Therefore, LSIs always have local scope.  They do not have
   any meaning outside the host running the application.  The ORCHID is
   used on the wire; not the LSI.  In the referral case, if it is not
   possible to rewrite the application level packets to use ORCHIDs
   instead of LSIs, it may be hard to make IPv4 referrals work in
   Internet-wide settings.  IPv4 LSIs have been successfully used in
   existing HIP deployments within a single corporate network.

3.  The HIP BONE Framework

   An overlay typically requires three types of operations:

   o  overlay maintenance.
   o  data storage and retrieval.
   o  connection management.

   Overlay maintenance operations deal with nodes joining and leaving
   the overlay and with the maintenance of the overlay's routing tables.
   Data storage and retrieval operations deal with nodes storing,

Camarillo, et al.        Expires April 29, 2010                 [Page 8]

Internet-Draft                  HIP BONE                    October 2009

   retrieving, and removing information in or from the overlay.
   Connection management operations deal with the establishment of
   connections and the exchange of lightweight messages among the nodes
   of the overlay, potentially in the presence of NATs.

   The HIP BONE framework uses HIP to perform connection management.
   Data storage and retrieval and overlay maintenance are to be
   implemented using protocols other than HIP.  For lack of a better
   name, these protocols are referred to as peer protocols.

   HIP BONE is a generic framework that allows the use of different peer
   protocols.  A particular HIP BONE instance uses a particular peer
   protocol.  The details on how to implement a HIP BONE using a given
   peer protocol need to be specified in a, so called, HIP BONE instance
   specification.  Section 3.5 discusses what details need to be
   specified by HIP BONE instance specifications.  For example, the HIP
   BONE instance specification for RELOAD [I-D.ietf-p2psip-base] is
   specified in [I-D.keranen-hip-reload-instance].

3.1.  Peer ID Assignment and Bootstrap

   Nodes in an overlay are primarily identified by their Peer IDs.
   (Note that the Peer ID concept here is a peer-layer protocol concept,
   distinct from the HIP-layer node identifiers.  Peer IDs may be long,
   may have some structure, and may consist of multiple parts.)
   Overlays typically have an enrollment server that can generate Peer
   IDs, or at least some part of the Peer ID, and sign certificates.  A
   certificate generated by an enrollment server authorizes a particular
   user to use a particular Peer ID in a particular overlay.  The way
   users and overlays are identified and the format for Peer IDs are
   defined by the peer protocol.

   The enrollment server of an overlay that were to use plain public
   keys as Peer IDs could just authorize users to use the public keys
   and HITs associated to their nodes.  This works well as long as the
   enrollment server is the one generating the public/private key pairs
   for all those nodes.  If the enrollment server authorizes users to
   use HITs that are generated directly by the nodes themselves, the
   system is open to a type of chosen-peer-ID attack.

   However, in some cases it is impractical to have the enrollment
   server generate public/private key pairs for devices.  In these
   cases, the enrollment server simply generates Peer IDs whose format
   is defined by the peer protocol used in the overlay.  Since HIP needs
   ORCHIDs (and not any type of Peer ID) to work, hosts in the overlay
   will transform their Peer IDs into ORCHIDs, for example, by taking a
   hash of the Peer IDs or taking a hash of the Peer ID and the public
   key.  That is a similar process to the one a host follows to generate

Camarillo, et al.        Expires April 29, 2010                 [Page 9]

Internet-Draft                  HIP BONE                    October 2009

   a HIT from a public key.  In such scenarios, each host will need a
   certificate (e.g., in their HIP base exchanges) provided by the
   enrollment server to prove that they are authorized to use a
   particular ORCHID in the overlay.  Depending on how the certificates
   are constructed, they typically also need to contain the host's self-
   generated public key.  Depending on how the Peer IDs and public keys
   are attributed, different scenarios become possible.  For example,
   the Peer IDs may be attributed to users, there may be user public key
   identifiers, and there may be separate host public key identifiers.
   Authorization certificates can be used to bind the different types of
   identifiers together.

   Bootstrap issues such as how to locate an enrollment or a bootstrap
   server belong to the peer protocol.

3.2.  Connection Establishment

   Nodes in an overlay need to establish connection with other nodes in
   different cases.  For example, a node typically has connections to
   the nodes in its forwarding table.  Nodes also need to establish
   connections with other nodes in order to exchange application-layer

   As discussed earlier, HIP uses the base exchange to establish
   connections.  A HIP endpoint (the initiator) initiates a HIP base
   exchange with a remote endpoint by sending an I1 packet.  The
   initiator sends the I1 packet to the remote endpoint's locator.
   Initiators that do not have any locator for the remote endpoint need
   to use a rendezvous service.  Traditionally, a HIP rendezvous server
   [RFC5204] has provided such a rendezvous service.  In HIP BONE, the
   overlay itself provides the rendezvous service.

   Therefore, in HIP BONE, a node uses an I1 packet (as usual) to
   establish a connection with another node in the overlay.  Nodes in
   the overlay forward I1 packets in a hop-by-hop fashion according to
   the overlay's routing table towards its destination.  This way, the
   overlay provides a rendezvous service between the nodes establishing
   the connection.  If the overlay nodes have active connections with
   other nodes in their forwarding tables and if those connections are
   protected (typically with IPsec ESP), I1 packets may be sent over
   protected connections between nodes.  Alternatively, if there is no
   such an active connection but the node forwarding the I1 packet has a
   valid locator for the next hop, the I1 packets may be forwarded
   directly, in a similar fashion to how I1 packets are today forwarded
   by a HIP rendezvous server.

   Since HIP supports NAT traversal, a HIP base exchange over the
   overlay will perform an ICE [I-D.ietf-mmusic-ice] offer/answer

Camarillo, et al.        Expires April 29, 2010                [Page 10]

Internet-Draft                  HIP BONE                    October 2009

   exchange between the nodes that are establishing the connection.  In
   order to perform this exchange, the nodes need to first gather
   candidate addresses.  Which nodes can be used to obtain reflexive
   address candidates and which ones can be used to obtain relayed
   candidates is defined by the peer protocol.

3.3.  Lightweight Message Exchanges

   In some cases, nodes need to perform a lightweight query to another
   node (e.g., a request followed by a single response).  In this
   situation, establishing a connection using the mechanisms in
   Section 3.2 for a simple query would be an overkill.  A better
   solution is to forward a HIP message through the overlay with the
   query and another one with the response to the query.  The payload of
   such HIP packets is integrity protected [I-D.nikander-hip-hiccups].
   Nodes in the overlay forward this HIP packet in a hop-by-hop fashion
   according to the overlay's routing table towards its destination,
   typically through the protected connections established between them.
   Again, the overlay acts as a rendezvous server between the nodes
   exchanging the messages.

3.4.  Participating in Multiple Overlays

   It is possible for a HIP host to participate simultaneously in
   multiple different overlay networks and a host needs to know to which
   overlay an incoming HIP message belongs to.  Thus all HIP messages
   that are sent via an overlay, or as a part of operations specific to
   a certain overlay, MUST contain an OVERLAY_ID parameter (shown in
   Figure 3) with the identifier of the corresponding overlay network.
   Instance specifications define how the identifier is generated for
   different types of overlay networks.  The generation mechanism SHOULD
   be such that it is unlikely to generate the same identifier for two
   different overlay instances and hence it is RECOMMENDED that the
   identifier contains at least 32 bits of randomness.

Camarillo, et al.        Expires April 29, 2010                [Page 11]

Internet-Draft                  HIP BONE                    October 2009

      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
     |             Type              |             Length            |
     |                           Identifier                          /
     /                                               |    Padding    |

     Type        [ TBD by IANA: 980 ]
     Length      Length of the Identifier in octets
     Identifier  The overlay identifier
     Padding     0-7 bytes of padding if the identifier length is not
                 multiple of 8 bytes

               Figure 3: Format of the OVERLAY_ID parameter

   OPEN ISSUE: this introduces normative language to the architecture
   draft.  Should this parameter be defined in some other (possibly
   stand alone) draft?

3.5.  HIP BONE Instantiation

   As discussed in Section 3, HIP BONE is a generic framework that
   allows using different peer protocols.  A particular HIP BONE
   instance uses a particular peer protocol.  The details on how to
   implement a HIP BONE using a given peer protocol need to be specified
   in a, so called, HIP BONE instance specification.  A HIP BONE
   instance specification needs to define, minimally:

   o  the peer protocol to be used.
   o  how to transform the peer IDs used by the peer protocol into the
      ORCHIDs that will be used in HIP.
   o  which peer protocol primitives trigger HIP messages.
   o  how the overlay identifier is generated.

   Additionally, a HIP BONE instance specification may need to specify
   other details that are specific to the peer protocol used.

   As an example, the HIP BONE instance specification for RELOAD
   [I-D.ietf-p2psip-base] is specified in

   It is assumed that areas not covered by a particular HIP BONE
   instance specification are specified by the peer protocol or
   elsewhere.  These areas include:

Camarillo, et al.        Expires April 29, 2010                [Page 12]

Internet-Draft                  HIP BONE                    October 2009

   o  the algorithm to create the overlay (e.g., a DHT).
   o  overlay maintenance functions.
   o  data storage and retrieval functions.
   o  format and structure of peer IDs.
   o  the process for obtaining a peer ID.
   o  bootstrap function
   o  how to select STUN and TURN servers for the candidate address
      collection process in NAT traversal scenarios.
   o  for what type of traffic or messages it is appropriate to use
      lightweight message exchanges.

   Note that the border between HIP BONE instance specification and a
   peer protocol specifications is blurry.  Depending on how generic the
   specification of a given peer protocol is, its associated HIP BONE
   instance specification may need to specify more or less details.
   Also, a HIP BONE instance specification may leave certain areas
   unspecified in order to leave their configuration up to each
   particular overlay.

4.  Advantages of Using HIP BONE

   Using HIP BONE, as opposed to a peer protocol, to perform connection
   management in an overlay has a set of advantages.  HIP BONE can be
   used by any peer protocol.  This keeps each peer protocol from
   defining primitives needed for connection management (e.g.,
   primitives to establish connections and to tunnel messages through
   the overlay) and NAT traversal.  Having this functionality at a lower
   layer allows multiple upper-layer protocols to take advantage of it.

   Additionally, having a solution that integrates mobility and
   multihoming is useful in many scenarios.  Peer protocols do not
   typically specify mobility and multihoming solutions.  Combining a
   peer protocol including NAT traversal with a separate mobility
   mechanism and a separate multihoming mechanism can easily lead to
   unexpected (and unpleasant) interactions.

5.  Architectural Considerations

   Architecturally, HIP can be considered to create a new thin "waist"
   layer on the top of the IPv4 and IPv6 networks; see Figure 4.  The
   HIP layer itself consists of the HIP signaling protocol and one or
   more data transport protocols; see Figure 5.  The HIP signaling
   packets and the data transport packets can take different routes.  In
   the HIP BONE, the HIP signaling packets are typically first routed
   through the overlay and then directly (if possible), while the data
   transport packets are typically routed only directly between the end

Camarillo, et al.        Expires April 29, 2010                [Page 13]

Internet-Draft                  HIP BONE                    October 2009


                  |    Transport (using HITs or LSIs)    |
                  |                 HIP                  |
                  |      IPv4        |       IPv6        |

                       Figure 4: HIP as a thin waist

                  |  HIP signaling   |  data transports  |

                       Figure 5: HIP layer structure

   In HIP BONE, the peer protocol creates a new signaling layer on the
   top of HIP.  It is used to set up forwarding paths for HIP signaling
   messages.  This is a similar relationship that an IP routing
   protocol, such as OSPF, has to the IP protocol itself.  In the HIP
   BONE case, the peer protocol plays a role similar to OSPF, and HIP
   plays a role similar to IP.  The ORCHIDs are used for forwarding HIP
   packets according to the information in the routing tables.  The peer
   protocols are used to exchange routing information based on Peer IDs
   and public keys, and to construct the routing tables.

   Architecturally, routing tables are located between the peer protocol
   and HIP, as shown in Figure 6.  The peer protocol constructs the
   routing table and keeps it updated.  The HIP layer accesses the
   routing table in order to make routing decisions.  The bootstrap of a
   HIP BONE overlay does not create circular dependencies between the
   peer protocol (which needs to use HIP to establish connections with
   other nodes) and HIP (which needs the peer protocol to know how to
   route messages to other nodes) for the same reasons as the bootstrap
   of an IP network does not create circular dependencies between OSPF
   and IP.  The first connections established by the peer protocol are
   with nodes whose locators are known.  HIP establishes those
   connections as any connection between two HIP nodes where no overlays
   are present.  That is, there is no need for the overlay to provide a
   rendezvous service for those connections.

Camarillo, et al.        Expires April 29, 2010                [Page 14]

Internet-Draft                  HIP BONE                    October 2009

                  |            Peer protocol             |
                  |            Routing table             |
                  |                 HIP                  |

                         Figure 6: Routing tables

   It is possible that different overlays use different routing table
   formats.  For example, the structure of the routing tables of two
   overlays based on different DHTs (Distributed Hash Tables) may be
   very different.  In order to make routing decisions, the HIP layer
   needs to convert the routing table generated by the peer protocol
   into a forwarding table that allows the HIP layer select a next-hop
   for any packet being routed.

   In HIP BONE, the HIP usage of public keys and deriving ORCHIDs
   through a hash function can be utilized at the peer protocol side to
   better secure routing table maintenance and to protect against
   chosen-peer-ID attacks.

   The HIP BONE provides quite a lot of flexibility with regards to how
   to arrange the different protocols in detail.  Figure 7 shows one
   potential stack structure.

                  | peer protocols        |     media    |
                  | HIP signaling    |   data transport  |
                  |                                      |
                  | NAT    | non-NAT |                   |
                  |                  |                   |
                  |      IPv4        |       IPv6        |

                Figure 7: Example HIP BONE stack structure

6.  Security Considerations

   This document provides a high-level framework to build HIP-based
   overlays.  The security properties of HIP and its extensions used in
   this framework are discussed in their respective specifications.
   Those security properties can be affected by the way HIP is used in a

Camarillo, et al.        Expires April 29, 2010                [Page 15]

Internet-Draft                  HIP BONE                    October 2009

   particular overlay (e.g., by how ORCHIDs are derived from Peer IDs).
   However, those properties are mostly affected by the design decisions
   made to build a particular overlay; not so much by how this high-
   level framework is specified in this document.  Such design decisions
   are typically documented in a HIP BONE instance specification, which
   will describe the security properties of the resulting overlay.

7.  Acknowledgements

   HIP BONE is based on ideas coming from conversations and discussions
   with a number of people in the HIP and P2PSIP communities.  In
   particular, Philip Matthews, Eric Cooper, Joakim Koskela, Thomas
   Henderson, Bruce Lowekamp, and Miika Komu provided useful input on

8.  IANA Considerations

   This document does not contain any IANA actions.

9.  References

9.1.  Normative References

   [RFC4843]  Nikander, P., Laganier, J., and F. Dupont, "An IPv6 Prefix
              for Overlay Routable Cryptographic Hash Identifiers
              (ORCHID)", RFC 4843, April 2007.

   [RFC5201]  Moskowitz, R., Nikander, P., Jokela, P., and T. Henderson,
              "Host Identity Protocol", RFC 5201, April 2008.

   [RFC5202]  Jokela, P., Moskowitz, R., and P. Nikander, "Using the
              Encapsulating Security Payload (ESP) Transport Format with
              the Host Identity Protocol (HIP)", RFC 5202, April 2008.

   [RFC5204]  Laganier, J. and L. Eggert, "Host Identity Protocol (HIP)
              Rendezvous Extension", RFC 5204, April 2008.

   [RFC5205]  Nikander, P. and J. Laganier, "Host Identity Protocol
              (HIP) Domain Name System (DNS) Extensions", RFC 5205,
              April 2008.

   [RFC5206]  Nikander, P., Henderson, T., Vogt, C., and J. Arkko, "End-
              Host Mobility and Multihoming with the Host Identity
              Protocol", RFC 5206, April 2008.

Camarillo, et al.        Expires April 29, 2010                [Page 16]

Internet-Draft                  HIP BONE                    October 2009

   [RFC5338]  Henderson, T., Nikander, P., and M. Komu, "Using the Host
              Identity Protocol with Legacy Applications", RFC 5338,
              September 2008.

              Komu, M. and T. Henderson, "Basic Socket Interface
              Extensions for Host Identity Protocol (HIP)",
              draft-ietf-hip-native-api-09 (work in progress),
              September 2009.

              Komu, M., Henderson, T., Tschofenig, H., Melen, J., and A.
              Keraenen, "Basic HIP Extensions for Traversal of Network
              Address Translators", draft-ietf-hip-nat-traversal-08
              (work in progress), June 2009.

              Nikander, P., Camarillo, G., and J. Melen, "HIP (Host
              Identity Protocol) Immediate Carriage and Conveyance of
              Upper-  layer Protocol Signaling (HICCUPS)",
              draft-nikander-hip-hiccups-04 (work in progress),
              August 2009.

9.2.  Informative References

   [RFC5389]  Rosenberg, J., Mahy, R., Matthews, P., and D. Wing,
              "Session Traversal Utilities for NAT (STUN)", RFC 5389,
              October 2008.

              Rosenberg, J., "Interactive Connectivity Establishment
              (ICE): A Protocol for Network Address  Translator (NAT)
              Traversal for Offer/Answer Protocols",
              draft-ietf-mmusic-ice-19 (work in progress), October 2007.

              Jennings, C., Lowekamp, B., Rescorla, E., Baset, S., and
              H. Schulzrinne, "REsource LOcation And Discovery (RELOAD)
              Base Protocol", draft-ietf-p2psip-base-04 (work in
              progress), October 2009.

              Keranen, A. and G. Camarillo, "HIP BONE Instance
              Specification for RELOAD",
              draft-keranen-hip-reload-instance-00 (work in progress),
              July 2009.

Camarillo, et al.        Expires April 29, 2010                [Page 17]

Internet-Draft                  HIP BONE                    October 2009

Authors' Addresses

   Gonzalo Camarillo
   Hirsalantie 11
   Jorvas  02420

   Email: Gonzalo.Camarillo@ericsson.com

   Pekka Nikander
   Hirsalantie 11
   Jorvas  02420

   Email: Pekka.Nikander@ericsson.com

   Jani Hautakorpi
   Hirsalantie 11
   Jorvas  02420

   Email: Jani.Hautakorpi@ericsson.com

   Ari Keranen
   Hirsalantie 11
   02420 Jorvas

   Email: Ari.Keranen@ericsson.com

   Alan Johnston
   St. Louis, MO  63124

   Email: alan@sipstation.com

Camarillo, et al.        Expires April 29, 2010                [Page 18]