Operational Guidance for IPv6 Deployment in IPv4 Sites using ISATAP
draft-templin-v6ops-isops-15.txt

Abstract

Many end user sites in the Internet today still have predominantly IPv4 internal infrastructures. These sites range in size from small home/office networks to large corporate enterprise networks, but share the commonality that IPv4 continues to provide satisfactory internal routing and addressing services for most applications. As more and more IPv6-only services are deployed in the Internet, however, end user devices within such sites will increasingly require at least basic IPv6 functionality for external access. This document therefore provides operational guidance for deployment of IPv6 within predominantly IPv4 sites using the Intra-Site Automatic Tunnel Addressing Protocol (ISATAP).

Status of this Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."

This Internet-Draft will expire on October 11, 2012.

Copyright Notice

Copyright (c) 2012 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

Table of Contents

• 1. Introduction
• 2. Enabling IPv6 Services using ISATAP
• 3. SLAAC Services
• 3.1. Advertising ISATAP Router Behavior
• 3.2. ISATAP Host Behavior
• 3.3. Reference Operational Scenario - Shared Prefix Model
• 3.4. Reference Operational Scenario - Individual Prefix Model
• 3.5. SLAAC Site Administration Guidance
• 3.6. Loop Avoidance
• 3.7. Interface Identifier Compatibility Considerations
• 4. Manual Configuration
• 5. Scaling Considerations
• 6. Site Renumbering Considerations
• 7. Path MTU Considerations
• 8. Alternative Approaches
• 9. IANA Considerations
• 10. Security Considerations
• 11. Acknowledgments
• 12. References
• 12.1. Normative References
• 12.2. Informative References
• Author's Address

1. Introduction

End user sites in the Internet today currently use IPv4 routing and addressing internally for core operating functions such as web browsing, filesharing, network printing, e-mail, teleconferencing and numerous other site-internal networking services. Such sites typically have an abundance of public or private IPv4 addresses for internal networking, and are separated from the public Internet by firewalls, packet filtering gateways, proxies, address translators and other site border demarcation devices. To date, such sites have had little incentive to enable IPv6 services internally [RFC1687].

End-user sites that currently use IPv4 services internally come in endless sizes and varieties. For example, a home network behind a Network Address Translator (NAT) may consist of a single link supporting a few laptops, printers etc. As a larger example, a small business may consist of one or a few offices with several networks connecting considerably larger numbers of computers, routers, handheld devices, printers, faxes, etc. Moving further up the scale, large banks, restaurants, major retailers, large corporations, etc. may consist of hundreds or thousands of branches worldwide that are tied together in a complex global enterprise network. Additional examples include personal-area networks, mobile vehicular networks, disaster relief networks, tactical military networks, and various forms of Mobile Ad-hoc Networks (MANETs). These cases and more are discussed in RANGERS[RFC6139].

With the proliferation of IPv6 devices in the public Internet, however, existing IPv4 sites will increasingly require a means for enabling IPv6 services so that hosts within the site can communicate with IPv6-only correspondents. Such services must be deployable with minimal configuration, and in a fashion that will not cause disruptions to existing IPv4 services. The Intra-Site Automatic Tunnel Addressing Protocol (ISATAP) [RFC5214] provides a simple-to-use service that sites can deploy in the near term to meet these requirements. This document therefore provides operational guidance for using ISATAP to enable IPv6 services within predominantly IPv4 sites while causing no disruptions to existing IPv4 services. The terminology of ISATAP (see: [RFC5214], Section 3) applies also to this document.

2. Enabling IPv6 Services using ISATAP

Existing sites within the Internet will soon need to enable IPv6 services. Larger sites typically obtain provider independent IPv6 prefixes from an Internet registry and advertise the prefixes into the IPv6 routing system on their own behalf, i.e., they act as an Internet Service Provider (ISP) unto themselves. Smaller sites that wish to enable IPv6 can arrange to obtain public IPv6 prefixes from an ISP, where the prefixes may be either purely native or the near-native prefixes offered by 6rd [RFC5969]. Alternatively, the site can obtain prefixes independently of an ISP e.g., via a tunnel broker [RFC3053], by using one of its public IPv4 addresses to form a 6to4 prefix [RFC3056][RFC3068], etc. (Note however that experience shows that the 6to4 method has some problems in current deployments that can lead to connectivity failures [RFC6343].) In any case, after obtaining IPv6 prefixes the site can automatically enable IPv6 services internally by configuring ISATAP.

The ISATAP service uses a Non-Broadcast, Multiple Access (NBMA) tunnel virtual interface model [RFC2491][RFC2529] based on IPv6-in-IPv4 encapsulation [RFC4213]. The encapsulation format can further use Differentiated Service (DS) [RFC2983] and Explicit Congestion Notification (ECN) [RFC3168] mapping between the inner and outer IP headers to ensure expected per-hop behavior within well-managed sites.

The ISATAP service is based on two node types known as advertising ISATAP routers and ISATAP hosts. (A third node type known as non-advertising ISATAP routers is defined in [I-D.templin-isupdate] but out of scope for this document.) Each node may further have multiple ISATAP interfaces (i.e., one interface for each site), and may act as an advertising ISATAP router on some of those interfaces and a simple ISATAP host on others. Hence, the node type is considered on a per-interface basis.

Advertising ISATAP routers configure their ISATAP interfaces as advertising router interfaces (see: [RFC4861], Section 6.2.2). ISATAP hosts configure their ISATAP interfaces as simple host interfaces and also coordinate their autoconfiguration operations with advertising ISATAP routers. In this sense, advertising ISATAP routers are "servers" while ISATAP hosts are "clients" in the service model.

Advertising ISATAP routers arrange to add their IPv4 address to the site's Potential Router List (PRL) so that ISATAP clients can discover them, as discussed in Sections 8.3.2 and 9 of [RFC5214]. Alternatively, site administrators could include IPv4 anycast addresses in the PRL and assign each such address to multiple advertising ISATAP routers. In that case, IPv4 routing within the site would direct the ISATAP client to the nearest advertising ISATAP router.

After the PRL is published, ISATAP clients within the site can automatically perform unicast IPv6 Neighbor Discovery Router Solicitation (RS) / Router Advertisement (RA) exchanges with advertising ISATAP routers using IPv6-in-IPv4 encapsulation [RFC4861][RFC5214]. In the exchange, the IPv4 source address of the RS and the destination address of the RA are an IPv4 address of the client, while the IPv4 destination address of the RS and the source address of the RA are an IPv4 address of the server found in the PRL. Similarly, the IPv6 source address of the RS is a link-local ISATAP address that embeds the client's IPv4 address, while the source address of the RA is a link-local ISATAP address that embeds the server's IPv4 address. (The destination addresses of the RS and RA may be either the neighbor's link-local ISATAP address or a link-scoped multicast address depending on the implementation.)

Following router discovery, ISATAP clients can configure and assign IPv6 addresses and/or prefixes using Stateless Address AutoConfiguration (SLAAC) [RFC4862][RFC5214]. While out of scope for this document, use of the Dynamic Host Configuration Protocol for IPv6 (DHCPv6) [RFC3315] is also possible when necessary updates to the ISATAP base specification are implemented [I-D.templin-isupdate].

3. SLAAC Services

Predominantly IPv4 sites can enable SLAAC services for ISATAP clients that need to communicate with IPv6 correspondents. SLAAC services are enabled using either the "shared" or "individual" prefix model. In the shared prefix model, all advertising ISATAP routers advertise a common prefix (e.g., 2001:db8::/64) to ISATAP clients within the site. In the individual prefix model, advertising ISATAP router advertise individual prefixes (e.g., 2001:db8:0:1::/64, 2001:db8:0:2::/64, 2001:db8:0:3::/64, etc.) to ISATAP clients within partition of the site. Note that combinations of the shared and individual prefix models are also possible, in which some of the site's ISATAP routers advertise shared prefixes and others advertise individual prefixes.

The following sections discuss operational considerations for enabling ISATAP SLAAC services within predominantly IPv4 sites.

3.1. Advertising ISATAP Router Behavior

Advertising ISATAP routers that support SLAAC services send RA messages in response to RS messages received on an advertising ISATAP interface. SLAAC services are enabled when advertising ISATAP routers advertise non-link-local IPv6 prefixes in Prefix Information Options (PIOs) with the A flag set to 1[RFC4861]. When there are multiple advertising ISATAP routers, the routers can advertise a shared IPv6 prefix or individual IPv6 prefixes.

3.2. ISATAP Host Behavior

ISATAP hosts resolve the PRL and send RS messages to obtain RA messages from an advertising ISATAP router. When the host receives RA messages, it uses SLAAC to configure IPv6 addresses from any advertised prefixes with the A flag set to 1 as specified in [RFC4862][RFC5214] then assigns the addresses to the ISATAP interface. The host also assigns any of the advertised prefixes with the L flag set to 1 to the ISATAP interface. (Note that the IPv6 link-local prefix fe80::/64 is always considered on-link on an ISATAP interface.)

3.3. Reference Operational Scenario - Shared Prefix Model

Figure 1 depicts an example ISATAP network topology for allowing hosts within a predominantly IPv4 site to configure ISATAP services using SLAAC with the shared prefix model. The example shows two advertising ISATAP routers ('A', 'B'), two ISATAP hosts ('C', 'D'), and an ordinary IPv6 host ('E') outside of the site in a typical deployment configuration. In this model, routers 'A' and 'B' both advertise the same (shared) IPv6 prefix 2001:db8::/64 into the IPv6 routing system, and also advertise the prefix to ISATAP clients within the site for SLAAC purposes.

                 .-(::::::::)      2001:db8:1::1
              .-(::: IPv6 :::)-.  +-------------+
             (:::: Internet ::::) | IPv6 Host E |
              `-(::::::::::::)-'  +-------------+
                 `-(::::::)-'
             ,~~~~~~~~~~~~~~~~~,
        ,----|companion gateway|--.
       /     '~~~~~~~~~~~~~~~~~'  :
      /                           |.
   ,-'                              `.
  ;  +------------+   +------------+  )
  :  |  Router A  |   |  Router B  |  /
   : |  (isatap)  |   |  (isatap)  |  :
   : | 192.0.2.1  |   | 192.0.2.1  | ;
   + +------------+   +------------+  \
  fe80::*:192.0.2.1   fe80::*:192.0.2.1
  | 2001:db8::/64       2001:db8::/64  |
  |                                   ;
  :              IPv4 Site         -+-'
   `-.       (PRL: 192.0.2.1)       .)
      \                           _)
       `-----+--------)----+'----'
  fe80::*:192.0.2.18          fe80::*:192.0.2.34
2001:db8::*:192.0.2.18      2001:db8::*:192.0.2.34
  +--------------+           +--------------+
  |   (isatap)   |           |   (isatap)   |
  |    Host C    |           |    Host D    |
  +--------------+           +--------------+

(* == "5efe")

Figure 1, advertising ISATAP routers 'A' and 'B' within the IPv4 site connect to the IPv6 Internet either directly or via a companion gateway. The routers advertise the shared prefix 2001:db8::/64 into the IPv6 Internet routing system either as a singleton /64 or as part of a shorter aggregated IPv6 prefix if the routing system will not accept prefixes as long as a /64. For the purpose of this example, we also assume that the IPv4 site is configured within multiple IPv4 subnets - each with an IPv4 prefix length of /28.

Advertising ISATAP routers 'A' and 'B' both configure the IPv4 anycast address 192.0.2.1 on a site-interior IPv4 interface, then configure an advertising ISATAP router interface for the site with link-local ISATAP address fe80::5efe:192.0.2.1. The site administrator then places the single IPv4 address 192.0.2.1 in the site's PRL. 'A' and 'B' then both advertise the anycast address/prefix into the site's IPv4 routing system so that ISATAP clients can locate the router that is topologically closest. (Note: advertising ISATAP routers can also use individual IPv4 unicast addresses instead of, or in addition to, a shared IPv4 anycast address. In that case, the PRL will contain multiple IPv4 addresses of advertising routers - some of which may be anycast and others unicast.)

ISATAP host 'C' connects to the site via an IPv4 interface with address 192.0.2.18/28, and also configures an ISATAP host interface with link-local ISATAP address fe80::5efe:192.0.2.18 over the IPv4 interface. 'C' next resolves the PRL, and sends an RS message to the IPv4 address 192.0.2.1, where IPv4 routing will direct it to the closest of either 'A' or 'B'. Assuming 'A' is closest, 'C' receives an RA from 'A' then configures a default IPv6 route with next-hop address fe80::5efe:192.0.2.1 via the ISATAP interface and processes the IPv6 prefix 2001:db8::/64 advertised in the PIO. If the A flag is set in the PIO, 'C' uses SLAAC to automatically configure the IPv6 address 2001:db8::5efe:192.0.2.18 (i.e., an address with an ISATAP interface identifier) and assigns it to the ISATAP interface. If the L flag is set, 'C' also assigns the prefix 2001:db8::/64 to the ISATAP interface, and the IPv6 address becomes a true ISATAP address.

In the same fashion, ISATAP host 'D' configures its IPv4 interface with address 192.0.2.34/28 and configures its ISATAP interface with link-local ISATAP address fe80::5efe:192.0.2.34. 'D' next performs an RS/RA exchange that is serviced by 'B', then uses SLAAC to autoconfigure the address 2001:db8::5efe:192.0.2.34 and a default IPv6 route with next-hop address fe80::5efe:192.0.2.1. Finally, IPv6 host 'E' connects to an IPv6 network outside of the site. 'E' configures its IPv6 interface in a manner specific to its attached IPv6 link, and autoconfigures the IPv6 address 2001:db8:1::1.

Following this autoconfiguration, when host 'C' inside the site has an IPv6 packet to send to host 'E' outside the site, it prepares the packet with source address 2001:db8::5efe:192.0.2.18 and destination address 2001:db8:1::1. 'C' then uses IPv6-in-IPv4 encapsulation to forward the packet to the IPv4 address 192.0.2.1 which will be directed to 'A' based on IPv4 routing. 'A' in turn decapsulates the packet and forwards it into the public IPv6 Internet where it will be conveyed to 'E' via normal IPv6 routing. In the same fashion, host 'D' uses IPv6-in-IPv4 encapsulation via its default router 'B' to send IPv6 packets to IPv6 Internet hosts such as 'E'.

When host 'E' outside the site sends IPv6 packets to ISATAP host 'C' inside the site, the IPv6 routing system may direct the packet to either of 'A' or 'B'. If the site is not partitioned internally, the router that receives the packet can use ISATAP to statelessly forward the packet directly to 'C'. If the site may be partitioned internally, however, the packet must first be forwarded to 'C's serving router based on IPv6 routing information. This implies that, in a partitioned site, the advertising ISATAP routers must connect within a full or partial mesh of IPv6 links, and must either run a dynamic IPv6 routing protocol or configure static routes so that incoming IPv6 packets can be forwarded to the correct serving router.

In this example, 'A' can configure the IPv6 route 2001:db8::5efe:192.0.2.32/124 with the IPv6 address of the next hop toward 'B' in the mesh network as the next hop, and 'B' can configure the IPv6 route 2001:db8::5efe:192.0.2.16/124 with the IPv6 address of the next hop toward 'A' as the next hop. (Notice that the /124 prefixes properly cover the /28 prefix of the IPv4 address that is embedded within the IPv6 address.) In that case, when 'A' receives a packet from the IPv6 Internet with destination address 2001:db8::5efe:192.0.2.34, it first forwards the packet toward 'B' over an IPv6 mesh link. 'B' in turn uses ISATAP to forward the packet into the site, where IPv4 routing will direct it to 'D'. In the same fashion, when 'B' receives a packet from the IPv6 Internet with destination address 2001:db8::5efe:192.0.2.18, it first forwards the packet toward 'A' over an IPv6 mesh link. 'A' then uses ISATAP to forward the packet into the site, where IPv4 routing will direct it to 'C'.

Finally, when host 'C' inside the site connects to host 'D' inside the site, it has the option of using the native IPv4 service or the ISATAP IPv6-in-IPv4 encapsulation service. When there is operational assurance that IPv4 services between the two hosts are available, the hosts may be better served to continue to use legacy IPv4 services in order to avoid encapsulation overhead and to avoid any IPv4 protocol-41 filtering middleboxes that may be in the path. If 'C' and 'D' may be in different IPv4 network partitions, however, IPv6-in-IPv4 encapsulation should be used with one or both of routers 'A' and 'B' serving as intermediate gateways.

3.4. Reference Operational Scenario - Individual Prefix Model

Figure 2 depicts an example ISATAP network topology for allowing hosts within a predominantly IPv4 site to configure ISATAP services using SLAAC with the individual prefix model. The example shows two advertising ISATAP routers ('A', 'B'), two ISATAP hosts ('C', 'D'), and an ordinary IPv6 host ('E') outside of the site in a typical deployment configuration. In the figure, ISATAP routers 'A' and 'B' both advertise different prefixes taken from the aggregated prefix 2001:db8::/48, with 'A' advertising 2001:db8:0:1::/64 and 'B' advertising 2001:db8:0:2::/64.

                 .-(::::::::)      2001:db8:1::1
              .-(::: IPv6 :::)-.  +-------------+
             (:::: Internet ::::) | IPv6 Host E |
              `-(::::::::::::)-'  +-------------+
                 `-(::::::)-'
             ,~~~~~~~~~~~~~~~~~,
        ,----|companion gateway|--.
       /     '~~~~~~~~~~~~~~~~~'  :
      /                           |.
   ,-'                              `.
  ;  +------------+   +------------+  )
  :  |  Router A  |   |  Router B  |  /
   : |  (isatap)  |   |  (isatap)  |  :
   : | 192.0.2.1  |   | 192.0.2.1  | ;
   + +------------+   +------------+  \
  fe80::*:192.0.2.1   fe80::*:192.0.2.1
  2001:db8:0:1::/64   2001:db8:0:2::/64
  |                                   ;
  :              IPv4 Site         -+-'
   `-.       (PRL: 192.0.2.1)       .)
      \                           _)
       `-----+--------)----+'----'
  fe80::*:192.0.2.18          fe80::*:192.0.2.34
2001:db8:0:1::*:192.0.2.18  2001:db8:0:2::*:192.0.2.34
  +--------------+           +--------------+
  |   (isatap)   |           |   (isatap)   |
  |    Host C    |           |    Host D    |
  +--------------+           +--------------+

(* == "5efe")

With reference to Figure 2, advertising ISATAP routers 'A' and 'B' within the IPv4 site connect to the IPv6 Internet either directly or via a companion gateway. Router 'A' advertises the individual prefix 2001:db8:0:1::/64 into the IPv6 Internet routing system, and router 'B' advertises the individual prefix 2001:db8:0:2::/64. The routers could instead both advertise a shorter shared prefix such as 2001:db8::/48 into the IPv6 routing system, but in that case they would need to configure a mesh of IPv6 links between themselves in the same fashion as described for the shared prefix model in Section 3.4. For the purpose of this example, we also assume that the IPv4 site is configured within multiple IPv4 subnets - each with an IPv4 prefix length of /28.

Advertising ISATAP routers 'A' and 'B' both configure the IPv4 anycast address 192.0.2.1 on a site-interior IPv4 interface, then configure an advertising ISATAP router interface for the site with link-local ISATAP address fe80::5efe:192.0.2.1. The site administrator then places the single IPv4 address 192.0.2.1 in the site's PRL. 'A' and 'B' then both advertise the anycast address/prefix into the site's IPv4 routing system so that ISATAP clients can locate the router that is topologically closest. (Note: advertising ISATAP routers can also use individual IPv4 unicast addresses instead of, or in addition to, a shared IPv4 anycast address. In that case, the PRL will contain multiple IPv4 addresses of advertising routers - some of which may be anycast and others unicast.)

ISATAP host 'C' connects to the site via an IPv4 interface with address 192.0.2.18/28, and also configures an ISATAP host interface with link-local ISATAP address fe80::5efe:192.0.2.18 over the IPv4 interface. 'C' next resolves the PRL, and sends an RS message to the IPv4 address 192.0.2.1, where IPv4 routing will direct it to the closest of either 'A' or 'B'. Assuming 'A' is closest, 'C' receives an RA from 'A' then configures a default IPv6 route with next-hop address fe80::5efe:192.0.2.1 via the ISATAP interface and processes the IPv6 prefix 2001:db8:0:1:/64 advertised in the PIO. If the A flag is set in the PIO, 'C' uses SLAAC to automatically configure the IPv6 address 2001:db8:0:1::5efe:192.0.2.18 (i.e., an address with an ISATAP interface identifier) and assigns it to the ISATAP interface. If the L flag is set, 'C' also assigns the prefix 2001:db8:0:1::/64 to the ISATAP interface, and the IPv6 address becomes a true ISATAP address.

In the same fashion, ISATAP host 'D' configures its IPv4 interface with address 192.0.2.34/28 and configures its ISATAP interface with link-local ISATAP address fe80::5efe:192.0.2.34. 'D' next performs an RS/RA exchange that is serviced by 'B', then uses SLAAC to autoconfigure the address 2001:db8:0:2::5efe:192.0.2.34 and a default IPv6 route with next-hop address fe80::5efe:192.0.2.1. Finally, IPv6 host 'E' connects to an IPv6 network outside of the site. 'E' configures its IPv6 interface in a manner specific to its attached IPv6 link, and autoconfigures the IPv6 address 2001:db8:1::1.

Following this autoconfiguration, when host 'C' inside the site has an IPv6 packet to send to host 'E' outside the site, it prepares the packet with source address 2001:db8::5efe:192.0.2.18 and destination address 2001:db8:1::1. 'C' then uses IPv6-in-IPv4 encapsulation to forward the packet to the IPv4 address 192.0.2.1 which will be directed to 'A' based on IPv4 routing. 'A' in turn decapsulates the packet and forwards it into the public IPv6 Internet where it will be conveyed to 'E' via normal IPv6 routing. In the same fashion, host 'D' uses IPv6-in-IPv4 encapsulation via its default router 'B' to send IPv6 packets to IPv6 Internet hosts such as 'E'.

When host 'E' outside the site sends IPv6 packets to ISATAP host 'C' inside the site, the IPv6 routing system will direct the packet to 'A' since 'A' advertises the individual prefix that matches 'C's destination address. 'A' can then use ISATAP to statelessly forward the packet directly to 'C'. If 'A' and 'B' both advertise the shared shorter prefix 2001:db8::/48 into the IPv6 routing system, however packets coming from 'E' may be directed to either 'A' or 'B'. In that case, the advertising ISATAP routers must connect within a full or partial mesh of IPv6 links the same as for the shared prefix model, and must either run a dynamic IPv6 routing protocol or configure static routes so that incoming IPv6 packets can be forwarded to the correct serving router.

In this example, 'A' can configure the IPv6 route 2001:db8:0:2::/64 with the IPv6 address of the next hop toward 'B' in the mesh network as the next hop, and 'B' can configure the IPv6 route 2001:db8:0.1::/64 with the IPv6 address of the next hop toward 'A' as the next hop. Then, when 'A' receives a packet from the IPv6 Internet with destination address 2001:db8:0:2::5efe:192.0.2.34, it first forwards the packet toward 'B' over an IPv6 mesh link. 'B' in turn uses ISATAP to forward the packet into the site, where IPv4 routing will direct it to 'D'. In the same fashion, when 'B' receives a packet from the IPv6 Internet with destination address 2001:db8:0:1::5efe:192.0.2.18, it first forwards the packet toward 'A' over an IPv6 mesh link. 'A' then uses ISATAP to forward the packet into the site, where IPv4 routing will direct it to 'C'.

Finally, when host 'C' inside the site connects to host 'D' inside the site, it has the option of using the native IPv4 service or the ISATAP IPv6-in-IPv4 encapsulation service. When there is operational assurance that IPv4 services between the two hosts are available, the hosts may be better served to continue to use legacy IPv4 services in order to avoid encapsulation overhead and to avoid any IPv4 protocol-41 filtering middleboxes that may be in the path. If 'C' and 'D' may be in different IPv4 network partitions, however, IPv6-in-IPv4 encapsulation should be used with one or both of routers 'A' and 'B' serving as intermediate gateways.

3.5. SLAAC Site Administration Guidance

In common practice, firewalls, gateways and packet filtering devices of various forms are often deployed in order to divide the site into separate partitions. In both the shared and individual prefix models described above, the entire site can be represented by the aggregate IPv6 prefix assigned to the site, while each site partition can be represented by "sliver" IPv6 prefixes taken from the aggregate. In order to provide a simple service that does not interact poorly with the site topology, site administrators should therefore institute an address plan to align IPv6 sliver prefixes with IPv4 site partition boundaries.

For example, in the shared prefix model in Section 3.3, the aggregate prefix is 2001:db8::/64, and the sliver prefixes are 2001:db8::5efe:192.0.2.0/124, 2001:db8::5efe:192.0.2.16/124, 2001:db8::5efe:192.0.2.32/124, etc. In the individual prefix model in Section 3.4, the aggregate prefix is 2001:db8::/48 and the sliver prefixes are 2001:db8:0:0::/64, 2001:db8:0:1::/64, 2001:db8:0:2::/64, etc.

When individual prefixes are used, site administrators can configure advertising ISATAP routers to advertise different individual prefixes to different sets of clients, e.g., based on the client's IPv4 subnet prefix such that the IPv6 prefixes are congruent with the IPv4 addressing plan. (For example, administrators can configure each advertising ISATAP router to provide services only to certain sets of ISATAP clients through inbound IPv6 Access Control List (ACL) entries that match the IPv4 subnet prefix embedded in the ISATAP interface identifier of the IPv6 source address). When a shared prefix is used, site administrators instead configure the ISATAP routers to advertise the shared prefix to all clients.

Advertising ISATAP routers can advertise prefixes with the (A, L) flags set to (1,0) so that ISATAP clients will use SLAAC to autoconfigure IPv6 addresses with ISATAP interface identifiers from the prefixes and assign them to the receiving ISATAP interface, but they will not assign the prefix itself to the ISATAP interface. In that case, the advertising router must assign the sliver prefix for the site partition to the advertising ISATAP interface. In this way, the advertising router considers the addresses covered by the sliver prefix as true ISATAP addresses, but the ISATAP clients themselves do not. This configuration enables a hub-and-spokes architecture which in some cases may be augmented by route optimization based on the receipt of ICMPv6 Redirects.

Site administrators can implement address selection policy rules [RFC3484] through explicit configurations in each ISATAP client. Site administrators implement this policy by configuring address selection policy rules [RFC3484] in each ISATAP client in order to give preference to IPv4 destination addresses over destination addresses derived from one of the client's IPv6 sliver prefixes.

For example, site administrators can configure each ISATAP client associated with a sliver prefix such as 2001:db8::5efe:192.0.2.64/124 to add the prefix to its address selection policy table with a lower precedence than the prefix ::ffff:0:0/96. In this way, IPv4 addresses are preferred over IPv6 addresses from within the same sliver. The prefix could be added to each ISATAP client either manually, or through an automated service such as a DHCP option [I-D.ietf-6man-addr-select-opt] discovered by the client, e.g., using Stateless DHCPv6 [RFC3736]. In this way, clients will use IPv4 communications to reach correspondents within the same IPv4 site partition, and will use IPv6 communications to reach correspondents in other partitions and/or outside of the site.

It should be noted that sliver prefixes longer than /64 cannot be advertised for SLAAC purposes. Also, sliver prefixes longer than /64 do not allow for interface identifier rewriting by address translators. These factors may favor the individual prefix model in some deployment scenarios, while the flexibility afforded by the shared prefix model may be more desirable in others. Additionally, if the network is small then the shared prefix model works well. If the network is large, however, a better alternative may be to deploy separate ISATAP routers in each region and have each advertise their own individual prefix.

Finally, site administrators should configure ISATAP routers to not send ICMPv6 Redirect messages to inform a source client of a better next hop toward the destination unless there is strong assurance that the client and the next hop are within the same IPv4 site partition.

3.6. Loop Avoidance

In sites that provide IPv6 services through ISATAP with SLAAC as described in this section, site administrators must take operational precautions to avoid routing loops. For example, each advertising ISATAP router should drop any incoming IPv6 packets that would be forwarded back to itself via another of the site's advertising routers. Additionally, each advertising ISATAP router should drop any encapsulated packets received from another advertising router that would be forwarded back to that same advertising router. This corresponds to the mitigation documented in Section 3.2.3 of [RFC6324], but other mitigations specified in that document can also be employed.

Note that IPv6 packets with link-local ISATAP addresses are exempt from these checks, since they cannot be forwarded by an IPv6 router and may be necessary for router-to-router coordinations.

3.7. Interface Identifier Compatibility Considerations

[RFC5214] Section 6.1 specifies the setting of the "u" bit in the Modified EUI-64 interface identifier format used by ISATAP. Implementations that comply with the specification set the "u" bit to 1 when the IPv4 address is known to be globally unique, however some legacy implementations unconditionally set the "u" bit to 0.

Implementations interpret the ISATAP interface identifier only within the link to which the corresponding ISATAP prefix is assigned, hence the value of the "u" bit is interpreted only within the context of an on-link prefix and not within a global context. Implementers are responsible for ensuring that their products are interoperable, therefore implementations must make provisions for ensuring "u" bit compatibility for intra-link communications.

Site administrators should accordingly configure access control list entries and other literal representations of ISATAP interface identifiers such that both values of the "u" bit are accepted. For example, if the site administrator configures an access control list entry that matches the prefix "fe80::0000:5efe:192.0.2.0/124" they should also configure a companion list entry that matches the prefix "fe80::0200:5efe:192.0.2.0/124.

4. Manual Configuration

When no autoconfiguration services are available (e.g., if there are no advertising ISATAP routers present), site administrators can use manual configuration to assign IPv6 addresses with ISATAP interface identifiers to the ISATAP interfaces of clients. Otherwise, site administrators should avoid manual configurations that would in any way invalidate the assumptions of the autoconfiguration service. For example, manually configured addresses may not be automatically renumbered during a site-wide renumbering event, which could subsequently result in communication failures.

5. Scaling Considerations

Section 3 depicts ISATAP network topologies with only two advertising ISATAP routers within the site. In order to support larger numbers of ISATAP clients (and/or multiple site partitions), the site can deploy more advertising ISATAP routers to support load balancing and generally shortest-path routing.

Such an arrangement requires that the advertising ISATAP routers participate in an IPv6 routing protocol instance so that IPv6 addresses/prefixes can be mapped to the correct ISATAP router. The routing protocol instance can be configured as either a full mesh topology involving all advertising ISATAP routers, or as a partial mesh topology with each advertising ISATAP router associating with one or more companion gateways. Each such companion gateway would in turn participate in a full mesh between all companion gateways.

6. Site Renumbering Considerations

Advertising ISATAP routers distribute IPv6 prefixes to ISATAP clients within the site. If the site subsequently reconnects to a different ISP, however, the site must renumber to use addresses derived from the new IPv6 prefixes [RFC1900][RFC4192][RFC5887].

For IPv6 services provided by SLAAC, site renumbering in the event of a change in an ISP-served IPv6 prefix entails a simple renumbering of IPv6 addresses and/or prefixes that are assigned to the ISATAP interfaces of clients within the site. In some cases, filtering rules (e.g., within site border firewall filtering tables) may also require renumbering, but this operation can be automated and limited to only one or a few administrative "touch points".

In order to renumber the ISATAP interfaces of clients within the site using SLAAC, advertising ISATAP routers need only schedule the services offered by the old ISP for deprecation and begin to advertise the IPv6 prefixes provided by the new ISP. ISATAP client interface address lifetimes will eventually expire, and the host will renumber its interfaces with addresses derived from the new prefixes. ISATAP clients should also eventually remove any deprecated SLAAC prefixes from their address selection policy tables, but this action is not time-critical.

Finally, site renumbering in the event of a change in an ISP-served IPv6 prefix further entails locating and rewriting all IPv6 addresses in naming services, databases, configuration files, packet filtering rules, documentation, etc. If the site has published the IPv6 addresses of any site-internal nodes within the public Internet DNS system, then the corresponding resource records will also need to be updated during the renumbering operation. This can be accomplished via secure dynamic updates to the DNS.

7. Path MTU Considerations

IPv6-in-IPv4 encapsulation overhead effectively reduces the size of IPv6 packets that can traverse the tunnel in relation to the actual Maximum Transmission Unit (MTU) of the underlying IPv4 network path between the encapsulator and decapsulator. Two methods for accommodating IPv6 path MTU discovery over IPv6-in-IPv4 tunnels (i.e., the static and dynamic methods) are documented in Section 3.2 of [RFC4213].

The static method places a "safe" upper bound on the size of IPv6 packets permitted to enter the tunnel, however the method can be overly conservative when larger IPv4 path MTUs are available. The dynamic method can accommodate much larger IPv6 packet sizes in some cases, but can fail silently if the underlying IPv4 network path does not return the necessary error messages.

This document notes that sites that include well-managed IPv4 links, routers and other network middleboxes are candidates for use of the dynamic MTU determination method, which may provide for a better operational IPv6 experience in the presence of IPv6-in-IPv4 tunnels. The dynamic MTU determination method can potentially also present a larger MTU to IPv6 correspondents outside of the site, since IPv6 path MTU discovery is considered robust even over the wide area in the public IPv6 Internet.

8. Alternative Approaches

[RFC4554] proposes a use of VLANs for IPv4-IPv6 coexistence in enterprise networks. The ISATAP approach provides a more flexible and broadly-applicable alternative, and with fewer administrative touch points.

The tunnel broker service [RFC3053] uses point-to-point tunnels that require end users to establish an explicit administrative configuration of the tunnel far end, which may be outside of the administrative boundaries of the site.

6to4 [RFC3056][RFC3068] and Teredo [RFC4380] provide "last resort" unmanaged automatic tunneling services when no other means for IPv6 connectivity is available. These services are given lower priority when the ISATAP managed service and/or native IPv6 services are enabled.

6rd [RFC5969] enables a stateless prefix delegation capability based on IPv4-embedded IPv6 prefixes, whereas ISATAP enables a stateful prefix delegation capability based on native IPv6 prefixes.

IRON [RFC6179], RANGER [RFC5720], VET [RFC5558] and SEAL [RFC5320] were developed as the "next-generation" of ISATAP and extend to a wide variety of use cases [RFC6139]. However, these technologies are not yet widely implemented or deployed.

9. IANA Considerations

This document has no IANA considerations.

10. Security Considerations

In addition to the security considerations documented in [RFC5214], sites that use ISATAP should take care to ensure that no routing loops are enabled [RFC6324]. Additional security concerns with IP tunneling are documented in [RFC6169].

11. Acknowledgments

The following are acknowledged for their insights that helped shape this work: Dmitry Anipko, Fred Baker, Brian Carpenter, Remi Despres, Thomas Henderson, Philip Homburg, Lee Howard, Ray Hunter, Joel Jaeggli, John Mann, Gabi Nakibly, Christopher Palmer, Hemant Singh, Mark Smith, Ole Troan, Gunter Van de Velde, ...

12. References

12.1. Normative References

12.2. Informative References

Author's Address