home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cis.ohio-state.edu!pacific.mps.ohio-state.edu!linac!uwm.edu!lll-winken!ptavv.llnl.gov!oberman
- From: oberman@ptavv.llnl.gov
- Newsgroups: comp.protocols.tcp-ip
- Subject: Re: What are the Pros/Cons of Multi-protocol routing?
- Message-ID: <1992Aug21.124812.1@ptavv.llnl.gov>
- Date: 21 Aug 92 20:48:12 GMT
- References: <86673@netnews.upenn.edu> <l9aeprINNsg0@noc.near.net>
- Sender: usenet@lll-winken.LLNL.GOV
- Distribution: usa
- Lines: 62
- Nntp-Posting-Host: ptavv.llnl.gov
-
- In article <l9aeprINNsg0@noc.near.net>, kwe@noc.near.net (Kent England) writes:
- > In article <86673@netnews.upenn.edu> Anthony Olejnik writes:
- >>
- >>Currently we have a pure IP only backbone.
- >>
- >>What are some of the advantages/disadvantages of keeping the
- >>backbone pure (with only IP)?
- >>
- >>What are the advantages/disadvantages of tunnelling the other
- >>protocols (AppleTalk/IPX/DecNet) within IP?
- >>
- >>Any comments would be *GREATLY* appreciated.
- >
- > TCP/IP has more sophisticated routing protocols than many of the other
- > internet protocols you can run, and it and OSI are the only protocols
- > suitable for a public internet.
- >
- > The disadvantage of a single protocol backbone is tunneling everything
- > else when routing is easier.
- >
- > But tunneling is sometimes necessary such as for getting AppleTalk across
- > the Internet, for source routed bridging across a complex WAN, or for
- > tunneling SDLC.
- >
- > Don't underestimate the complexity of supporting additional protocols
- > whether tunneled or routed. Make sure you have enough features and
- > filters to control them. But if they are important enough to your
- > situation, then by all means support them.
-
- I have a problem with oversimplifying anything. And this is way oversimplified.
-
- First, how big a backbone is involved? For a small backbone of a couple
- thousand systems, routing all of the mentioned protocols is no problem.
- AppleTalk and IPX don't scale well, though, for big nets. And all routed
- protocols must be centrally administered to avoid addressing conflicts.
-
- AppleTalk should be AppleTalk Phase 2. Phase 1 sends out to much overhead on a
- large net. And don't even thing about large scale tunnelling of AppleTalk. The
- RTMP will kill you. Routing can work to a limited extent, though.
-
- DECnet is quite suitable for wide area routing. It was designed for it and is
- generally much easier to manage than even IP. Many regionals and two of the
- three national backbones in the US carry DECnet quite well with a global DECnet
- internet of at least 20K nodes. (It's probably much larger, but it's hard to
- keep track.) But to join the global DECnet you must get addresses from a
- central authority, just like with IP.
-
- I guess I'd say the multi-protocol routing is much better than tunnelling in
- most cases. But you have to understand the protocols and how the operate on
- things like slow serial lines and with slow propagation. IP, DECnet, and OSI
- are well suited to large networks. AppleTalk and IPX are OK for smaller nets,
- but not to anything like the size of the Internet.
-
- I might also mention that DECnet has only 16 bits of address, so you have to
- deal with that when the net gets big. IP will have this problem shortly. At up
- to 160 bits, OSI is OK for a while.
-
- R. Kevin Oberman Lawrence Livermore National Laboratory
- Internet: koberman@llnl.gov (510) 422-6955
-
- Disclaimer: Don't take this too seriously. I just like to improve my typing
- and probably don't really know anything useful about anything.
-