Internet Engineering Task Force Arifumi Matsumoto
Internet-Draft Tomohiro Fujisaki
Expires: April 12, 2005 Hirotaka Matsuoka
Jun-ya Kato
NTT PF Lab.
October 12, 2004
Source Address Selection Policy Distribution for Multihoming
draft-arifumi-multi6-sas-policy-dist-00.txt
Status of this Memo
This document is an Internet-Draft and is subject to all provisions
of section 3 of RFC 3667. By submitting this Internet-Draft, we
certify that any applicable patent or other IPR claims of which we
are aware have been disclosed, or will be disclosed, and any of which
we become aware will be disclosed, in accordance with RFC 3668.
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 made obsolete 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.
Copyright Notice
Copyright (C) The Internet Society (2004). All Rights Reserved.
Abstract
This document describes a method for the distribution of source
address selection policy from ISPs to gateway routers for consumers
and from the gateways to end nodes. This method is particularly
effective when a consumer site has multiple address blocks. Every
end node is guided by the policy in selecting an appropriate source
address for each destination address and every gateway is guided by
Arifumi [Page 1]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
the policy in forwarding packets to appropriate next-hop ISPs. This
makes it possible for an end node to set a connection up without
being concerned about failures of transfer due to ingress filtering
by the ISPs, for ISP operators to manage consumers' behavior and
networking policy, and for consumers to be provided with networks
that are almost automatically robust and reliable.
1. Introduction
An IPv6 multihoming site has multiple nodes, each of which is
assigned multiple IPv6 addresses by up-stream ISPs. When there are
multiple up-stream ISPs, the means of selection of the ISP for an
outgoing packet is currently based on the destination address. In
general, however, each packet should have a source address that has
been allocated by the selected up-stream ISP. This is because the
routers of ISPs may be configured to perform ingress filtering with
the aim of blocking packets that have strange source addresses.
In this document, we propose a technique that is used both to
distribute policy information for source address selection at end
nodes and to establish a method for the forwarding of packets by
routers. This enables the control of incoming traffic from customer
sites by ISPs, the selection of appropriate source addresses by end
nodes, and the selection of outgoing ISPs in a way that is almost
certain to produce successful connection setup.
2. Concepts
In this document, we propose a method by which an ISP can distribute
source address selection policies to each end node at a customer
site. The policy information is particularly helpful to hosts in
which multihoming is used, since an end node can use the destination
address to select a source address that leads to a high probability
of successful setup for the connection.
An up-stream ISP is expected to use DHCPv6 Prefix Options [3633] to
delegate a certain portion of the IPv6 address space to its
subscribers. We propose a DHCPv6 new option, which contains a per-
delegating-prefix address-selection policy. By making use of this
option, an ISP can inform its customers of an address block that can
be reached through the ISP and of a corresponding source address of
packets, that is, a source address that must be used to reach the
given block. This is simply achieved through delegation of the
delegated source-address prefix and policy by the ISP.
The gateway router of a customer's site receives the delegated prefix
and address-selection policy mentioned above from its up-stream ISPs.
Arifumi [Page 2]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
The router in turn distributes this information to end nodes at the
site. Here, we propose an extension to the ND6 Router Advertisement
Message [2461] and a DHCPv6 [3315] new option to cover delivery of
address-selection policy to the end nodes.
The address-selection policy delivered to an end node is stored in
the form of a Policy Table as defined in RFC 3484 [3484].
Once the above series of processes is complete, an end node can
select an appropriate source address for a given destination address.
Routing of an outgoing packet to the corresponding up-stream ISP is
mainly guided by the source address of the packet; this avoids
blocking of the packet by ingress filtering.
This mechanism is particularly effective when a site subscribes to an
ISP or VPN service that provides connectivity to a certain closed
network as well as acting as an ISP for global network connectivity.
This is because selecting an appropriate source address for a given
destination address is crucial in such a network environment.
This approach gives end nodes an advance measure against connection
setup failure. At the same time, an ISP can control incoming traffic
from customers' sites, and the network managers of customers' sites
can reflect their networking policy to some extent by configuring
DHCPv6 or ND6 RA settings on routers. The last but not the least
significant feature to note here is that this sequence of passing,
processing, generation, and reflection of policy information can be
made almost totally automatic from the viewpoint of customers.
In the context of Multi6 WG activity, IMHO this kind of information
will be very useful in terms of efficiently and effectively setting
up connections no matter what modifications have been applied to the
protocol stacks of end nodes. Also, this will be very helpful for
those end nodes that haven't received the protocol-stack
modifications required for this technique, in other words, to legacy
nodes.
This document also serves as a proposal for the Multi6 WG to consider
a mechanism of the kind described as one element of the component
class.
Arifumi [Page 3]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
3. Overview
3.1 Multihome Site with Global-Closed Mixed Connectivity
==============
| Internet |
==============
|
2001:db8::/32 | 3ffe:1800::/32
+----+-+ +-+----+
| ISP1 | | ISP2 | (Closed Network)
+----+-+ +-+----+
| |
2001:db8:a::/48 | | 3ffe:1800:a::/48
(DHCP-PD') ++-------++ (DHCP-PD')
| Gateway |
+----+----+
| 2001:db8:a:1::/64
| 3ffe:1800:a:1::/64
| (RA'/DHCP')
------+---+----------
|
+-+----+ 2001:db8:a:1:[EUI64]
| Host | 3ffe:1800:a:1:[EUI64]
+------+
[Fig. 1]
Fig. 1 shows a multihome site that subscribes to two ISPs. One ISP
provides global network connectivity and the other provides
connectivity to a closed network but not to the Internet. This site
has one border router, labeled Gateway here, and the router may be
connected to up-stream ISPs through a physical or logical link, say
PPPoE or an IPsec Tunnel.
3.1.1 Description of Each Element
i) ISP -> Gateway
This figure shows that ISP1 has been allocated 2001:db8::/32 and
ISP2 has been allocated 3ffe:1800::/32. Each ISP delegates part of
its address block, called the "provider aggregatable (PA)" block,
to this customer site. Here, ISP1 and ISP2 use DHCP-PD to delegate
2001:db8:a::/48 and 3ffe:1800:a::/48, respectively.
In this document, we propose an extension to DHCP-PD, which is
actually a new DHCP option and called DHCP-PD' here. DHCP-PD'
gives DHCP servers (ISPs) functionality for delivering an address-
Arifumi [Page 4]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
selection policy in combination with a delegated prefix to a client
(gateway). In this example, ISP2 includes
Address Addr. Sel. Policy
3ffe:1800:a::/48 ---- 3ffe:1800::/32
in the PD and address-selection policy options sent to the client.
This means that the subscribers of ISP2 should use an address from
the delegated range, that is, 3ffe:1800:a::/48, when communicating
with 3ffe:1800::/32.
Selection of an appropriate source address is very important, and
this is particularly so when one of the subscribing networks is
closed as in this example. This is simply because a return packet
from the closed network can't possibly reach the session-
originating host if the return packet is destined for an address
beyond the range available to the closed ISP.
ISP1 also uses DHCP-PD', in this case to deliver its address-
selection policy to its customers. As ISP1 provides global network
connectivity, the PD and policy options will take the following
form.
Address Addr-Sel. Policy
2001:db8:a::/48 --+-- 2001:db8::/32
+-- ::/0 (all address)
This means that ISP1 can provide connectivity to 2001:db8::/32 and
act as a transit point for any other address (::/0) in the Internet
as long as the source address is that delegated by ISP1, namely
2001:db8:a::/48.
With regard to backward compatibility, a normal DHCP-PD packet,
which does not carry address-selection policy information of the
above type, should be deemed to have ::/0 as its policy field.
ii) Gateway -> Host
A gateway router receives address-delegation information and
address-selection policy from up-stream ISPs, in turn delivering
both to down-stream nodes. In this document, we propose DHCP new
option and an extension to RA (Router Advertisement Message). We
refer to these as DHCP' and RA', respectively. The gateway router
combines information given by multiple up-stream ISPs and
distributes the following information down-stream through DHCP' or
RA'.
Arifumi [Page 5]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
Address Addr. Sel. Policy
1 2001:db8:a:1::/64 --+-- 2001:db8::/32
+-- ::/0
2 3ffe:1800:a:1::/64 ----- 3ffe:1800::/32
This is just the combination of the information from the two up-
stream elements in the previous section, except that each
advertising address prefix is 64 bits long.
iii) Host
When a host receives an RA' or DHCP' message from the site gateway,
it configures addresses for each receiving network interface and
reflects address-selection policy in its RFC3484-based policy
table.
In this example, we propose that the policy table should be
configured as follows, by making use of the Label field defined in
RFC3484, and the relation between address prefix and address-
selection policy should be kept in this table.
Prefix Pref. Label
2001:db8::/32 10 100
::/0 10 100
2001:db8:a:1::/64 10 100
3ffe:1800::/32 10 200
3ffe:1800:a:1::/64 10 200
When this host sends a packet to, for example, 3ffe:1800:a::100,
whose longest matching entry in this table is 3ffe:1800::/32, the
host chooses the address beginning with 3ffe:1800:a:1:: as the
source address. The source-address selection algorithm will select
the longest entry that is a candidate source-address range and has
the same label as the longest matching address for the destination
address. In the same way, the source address of a packet destined
for 2001:db8::/32 or 2002::/16 will be 2001:db8:a:1:[EUI64].
Preference values are only used in the selection of destination
addresses. This document does not include an algorithm for
determining preference values.
iv) Gateway
As well as delivering addresses and policy information to hosts
through RA' or DHCP', the gateway forwards packets according to
policy information distributed by up-stream ISPs. One way of
implementing such forwarding or routing, called policy routing, is
based on the source addresses of out-going packets. Policy routing
Arifumi [Page 6]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
is illustrated in the table below.
Src. Next Hop
2001:db8:a:1::/64 ISP1
3ffe:1800:a:1::/64 ISP2
In this example, the next-hop is deterministically selected by the
destination address. So, normal destination-address-based routing
with the following routing table is sufficient.
Dst. Next Hop
2001:db8::/32 ISP1
::/0 ISP1
3ffe:1800::/32 ISP2
3.1.2 Discussion
The benefits of this scheme are very clear. Every end node can
determine which source address should be used and can send packets
without a risk of failure due to ingress filtering or the limited
reachability of a closed network.
What should be discussed from here is the need for and implementation
of policy enforcement to end nodes and the process of combining
multiple address-selection policies. It's so hard to combine two
policies automatically when a policy coming from an ISP conflicts
with another ISP's policy. We may also have to think about combining
or pruning algorithm to contain too much policy information in one
packet.
Arifumi [Page 7]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
3.2 Host with Multiple Home Addresses and Connectivity to Two Global
Networks
==================
| Internet |
==================
| |
2001:db8::/32 | | 3ffe:1800::/32
+----+-+ +-+----+
| ISP1 | | ISP2 |
+----+-+ +-+----+
| |
2001:db8:a::/48 | | 3ffe:1800:a::/48
(DHCP-PD') ++-------++ (DHCP-PD')
| Gateway |
+----+----+
| 2001:db8:a:1::/64
| 3ffe:1800:a:1::/64
| (RA'/DHCP')
------+---+----------
|
+-+----+ 2001:db8:a:1:[EUI64]
| Host | 3ffe:1800:a:1:[EUI64]
+------+
[Fig. 2]
Fig. 2 shows a host with multiple home addresses that subscribes to
two ISPs for connectivity to the Internet. The manner of address
delegation and allocation is as described in 3.1.
3.2.1 Description of Each Element
i) ISP -> Gateway
The difference between this and the previous example is that ISP2
provides global network connectivity, so the DHCP-PD' address-
selection policy option for ISP2 includes an additional entry.
Address Addr. Sel. Policy
3ffe:1800:a::/48 --+-- 3ffe:1800::/32
+--- ::/0
ii) Gateway -> Host
As both ISPs provide global network connectivity, the policy for
address-selection from the gateway router to the end nodes is of
the form shown below.
Arifumi [Page 8]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
Address Addr. Sel. Policy
1 2001:db8:a:1::/64 --+-- 2001:db8::/32
+-- ::/0
2 3ffe:1800:a:1::/64 --+-- 3ffe:1800::/32
+--- ::/0
iii) Host
Each end node will have the following address-selection policy
table.
Prefix. Pref. Label
2001:db8::/32 10 100
2001:db8:a:1::/64 10 100
(::/0 10 100)
3ffe:1800::/32 10 200
3ffe:1800:a:1::/64 10 200
(::/0 10 200)
Note that the end nodes are notified of an address-selection policy
that includes prefix ::/0 by both ISPs, hence a specific source
address for ::/0 can't be determined in the Label-Rule judgment
phase described in RFC3484. So, these entries for prefix ::/0 won't
actually be stored in the policy table, and this policy table won't
have any effect on source-address selection for packets that match
::/0. The source address in these cases will be determined by
following rules listed in RFC3484, such as longest match with the
destination address.
iv) Gateway
Unlike the previous example (3.1.1(iv)), normal destination-
address-based routing doesn't specify a particular next-hop.
Dst Next-Hop
2001:db8::/32 ISP1
::/0 ISP1
3ffe:1800::/32 ISP2
::/0 ISP2
So, source-address-based routing as described in 3.1.1(iv) will be
effective.
3.2.2 Discussion
One of the benefits of having an ISP provide address-selection policy
to its customers is that it can explicitly check incoming packets to
Arifumi [Page 9]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
see if it delegated their source addresses. Delivery of an address-
selection policy makes the following mechanisms possible.
- Since end nodes and routers in a multihoming site are given some
kind of routing information, they can select the route expected to
be optimal. In the above example, end nodes can communicate with
servers of the ISPs without any circumvention.
- Another possible usage of this framework is notification of
security policy. ISPs commonly apply IP-address-based filtering to
packets that attempt access to their services, such as POP, SMTP
and Web content, partly for security reasons and partly as a value-
added service for their customers.
IMHO one issue to be discussed is multipath routing. In the second of
the examples presented above, the router and host have two default
routes and thus have the potential to apply multipath techniques
[2991][2992]. To take greater advantage of multihoming, we should
probably consider using a multipath routing algorithm.
3.3 A Host Directly Connected to Multiple ISPs
==================
| Internet |
==================
| |
2001:db8::/32 | | 3ffe:1800::/32
| |
2001:db8:a::/48 | | 3ffe:1800:a::/48
(DHCP-PD') +----+-+ +-+----+ (DHCP-PD')
| GW1 | | GW2 |
+----+-+ +-+----+
2001:db8:a:1::/64 | | 3ffe:1800:a:1::/64
(RA'/DHCP') | | (RA'/DHCP')
-----+-+- -+-+-----
| |
2001:db8:a:1:[EUI64]++---++ 3ffe:1800:a:1:[EUI64]
| Host|
+-----+
[Fig. 3]
This example shows an end node directly connected to two ISPs, both
of which provide global network connectivity. The only difference
between this case and the case described in 3.2 is that the host
itself has to take on the role of gateway router, that is, of
applying routing policy based on source addresses.
Arifumi [Page 10]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
4. Security Considerations
With regard to the possibility of traffic abduction through the
announcement of a bogus policy, this scheme seems to neither lower
nor raise the security level obtained by the existing base-protocols,
such as DHCP-PD, DHCP and RA. However, it does raise the possibility
of a new form of DoS attack on routers and hosts, in which large
numbers of address-selection policies are generated by different
source addresses. We will have to discuss this and take precautionary
measures in designing the protocol specification.
5. Normative References
[3633] O. Troan and R. Droms, "IPv6 Prefix Options for Dynamic Host
Configuration Protocol (DHCP) version 6," RFC 3633, Dec. 2003.
[2461] T. Narten, E. Nordmark, and W. Simpson, "Neighbor Discovery
for IP Version 6 (IPv6)," RFC 2461, Dec. 1998.
[3315] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C. and M.
Carney, "Dynamic Host Configuration Protocol for IPv6
(DHCPv6)," RFC 3315, Jul. 2003.
[3484] R. Draves, "Default Address Selection for Internet Protocol
version 6 (IPv6)," RFC 3484, Feb. 2003.
[2991] D. Thaler and C. Hopps, "Multipath Issues in Unicast and
Multicast Next-Hop Selection," RFC 2991, Nov. 2000.
[2992] C. Hopps, "Analysis of an Equal-Cost Multi-Path Algorithm,"
RFC 2992, Nov. 2000.
Authors' Addresses
Arifumi Matsumoto
Nippon Telegraph and Telephone Corporation
Information Sharing Platform Laboratories
3-9-11 Midori-cho
Musashino-shi, Tokyo 180-8585 Japan
Phone: +81-422-59-3334
E-Mail: matsumoto.arifumi@lab.ntt.co.jp
Tomohiro Fujisaki
Nippon Telegraph and Telephone Corporation
Information Sharing Platform Laboratories
3-9-11 Midori-cho
Arifumi [Page 11]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
Musashino-shi, Tokyo 180-8585 Japan
Phone: +81-422-59-7351
E-Mail: fujisaki.tomohiro@lab.ntt.co.jp
Hirotaka Matsuoka
Nippon Telegraph and Telephone Corporation
Information Sharing Platform Laboratories
3-9-11 Midori-cho
Musashino-shi, Tokyo 180-8585 Japan
Phone: +81-422-59-4949
E-Mail: matsuoka.hirotaka@lab.ntt.co.jp
Jun-ya Kato
Nippon Telegraph and Telephone Corporation
Information Sharing Platform Laboratories
3-9-11 Midori-cho
Musashino-shi, Tokyo 180-8585 Japan
Phone: +81-422-59-2939
E-Mail: kato.junya@lab.ntt.co.jp
Full Copyright Statement
Copyright (C) The Internet Society (2004). 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 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
Arifumi [Page 12]
draft-arifumi-multi6-sas-policy-dist-00.txt October 2004
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.
Arifumi [Page 13]