Welcome to the UniDirectional Link Routing
home page

The problem

The Internet is mostly composed of networks interconnected via heterogeneous, but bidirectional and symmetric links. Heterogeneity means that links may have different bandwidth. Symmetry means that links have the same characteristics in each direction. Internet routing and transmission control protocols have been designed to optimize the transmission assuming bidirectional symmetric communication links.

Recently, new transmission media are proposed to provide higher bandwidth to the Internet. Satellite links with receive only antennas and cables offer respectively unidirectional and asymmetric access to the Internet are proposed. Both offer high speed bandwidth downstream, with zero bandwidth upstream for the satellite link and a lower bandwidth upstream for cables. Broadcast satellite links may therefore be used to provide "alternative" high speed receive only access to the Internet, with the return trafic sent through a low speed modem access. There is a need to integrate both unidirectional and asymmetric links in the Internet transparently. Unidirectional links are a priori a special case of asymmetric links.

However, there are specific problems to be solved in the case of unidirectional links (e.g. dynamic routing and transmission control). Three cases for unidirectional and/or asymmetric links may be envisaged:

1. unidirectional links on top of bidirectional underlying network (wired Internet)

2. bidirectional islands connected via unidirectional links.

3. the general case of asymmetric and possibly unidirectional links

In the UniDirectional Link Routing working group we focus on the "short term" (case 1) i.e. the support of alternative unidirectional link on top of a bidirectional internetwork. We do not consider the case of bidirectional islands connected via unidirectional links nor the general case of asymmetric and possibly unidirectional links. In addition, we focus on routing related problems. Therefore, the impact of unidirectional links on transport level protocols (TCP, RSVP) is out of scope for the working group.

Common Internet routing protocols (e.g. RIP, OSPF and DVMRP) can nolonger work properly because of the assumed links bidirectionality and symmetry. Concerning the short term case, there are two proposed approaches. The first is based on the modification of the common routing protocols (RIP, OSPF, DVMRP) in order to support unidirectional links. The second is to add a layer between the network interface and the routing software to emulate bi-directional links through tunnels. This second approach consists in masking the underlying unidirectionality with a tunneling mechanism under the  network level.

The UDLR working group

The udlr working group at the Internet Ingineering Task Force (IETF) focuses on the discussion of these two "short term" solutions for the support of a unidirectional link on top of a bidirectional internetwork (case 1). However, any idea concerning the case 2 or case 3 (see above section) is welcomed on the udlr mailing list <udlr@sophia.inria.fr>.

The purpose of the UDLR Working Group, is to study these approaches and suggest a short term solution that provides dynamic routing (including multicast) in the presence of unidirectional links.

For more details on the UDLR working group (chair, goals and milstones), the charter can be viewed directly from the IETF site.

Two UDLR BoF sessions started before becoming a Working Group:

1. 36th IETF meeting in Montreal (Quebec, CANADA), June 1996. Here are the minutes.

2. 37th IETF meeting in San Jose (California, U.S.), December 1996. Here are the minutes.

Here are the UDLR Working Groups which succeeded to the udlr BoF sessions (under construction):

1. 38th IETF meeting in Memphis (Tennessee, U.S.), April 1997. Here are the minutes.
2. 39th IETF meeting in Munich (Germany), August 1997. Here are the minutes.

Participating to the UDLR working group can be done via the UDLR mailing list:

• To subscribe, send a mail without content to: udlr-subscribe@sophia.inria.fr
• To unsubscribe, send a mail without content to: udlr-unsubscribe@sophia.inria.fr
• General discussion: udlr@sophia.inria.fr
• Archive: http://www-sop.inria.fr/rodeo/udlr/archive/threads.html
• If you have any questions regarding to the mailing list: udlr-owner@sophia.inria.fr

Documents

Drafts:

1. draft-ietf-udlr-lltunnel-04.txt
   A Link Layer Tunneling Mechanism for Unidirectional Links, E. Duros, W. Dabbous (INRIA Sophia-Antipolis), H. Izumiyama, N Fujii (WIDE), Y Zhang (HRL), February 1999

2. draft-udlr-vipre-00.txt
   VIPRE -- Virtual Internet Packet Relay, James Stepanek and Stephen Schwab (The Aerospace Corporation), Febreary 1997

3. draft-udlr-wide-tunnel-00.txt
   An IP tunneling approach for Uni-directional Link routing, Hidetaka Izumiyama, Akihiro Tosaka and Akira Kato (WIDE project). July 1997.

4. draft-udlr-dtpc-00.txt
   Dynamic Tunneling Path Configuration for Uni-directional Link Routing, Hidetaka Izumiyama and Akihiro Tosaka (WIDE project). July 1997.

5. draft-udlr-general-00.txt
   Supporting Unidirectional Paths in the Internet, Emmanuel Duros and Walid Dabbous, INRIA. June 1996.

6. draft-udlr-rip-00.txt
   Handling of unidirectional links with RIP, E. Duros and C. Huitema, INRIA, March 1996.

7. draft-udlr-ospf-00.txt
   Handling of unidirectional links with OSPF, E. Duros, INRIA, March 1996.

8. draft-udlr-dvmrp-00.txt
   Handling of unidirectional links with DVMRP, E. Duros, INRIA, March 1996.

The list is not exhaustive, so please let us know if you want to add a link to it. We would like to list here most of the links related to UDLR.

Questions, comments, suggestions ?

If you have any, please contact either to E. Duros or W. Dabbous.

E. Duros - W. Dabbous INRIA Sophia-Antipolis