home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.lans.ethernet
- Path: sparky!uunet!paladin.american.edu!howland.reston.ans.net!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!unixhub!unixhub.SLAC.Stanford.EDU!streater
- From: streater@unixhub.SLAC.Stanford.EDU (Tim Streater)
- Subject: Re: Dec and IEEE Spanning Tree differences
- Message-ID: <C03FEt.CMv@unixhub.SLAC.Stanford.EDU>
- Sender: news@unixhub.SLAC.Stanford.EDU
- Organization: Stanford Linear Accelerator Center
- References: <C01pny.4Kp@unixhub.SLAC.Stanford.EDU> <1992Dec30.023337.17569@netcom.com>
- Date: Wed, 30 Dec 1992 22:20:53 GMT
- Lines: 37
-
- Thanks to those who responded about this; I think I have the information I need.
-
- A word about why I might want to use more than one algorithm. We have a
- subnetted LAN here, with some of the subnets being *very* critical, even to
- the point that anything other than a minor intervention would need to wait
- several months before being implemented. Such subnets will typically have a
- redundant path *within* the subnet, implemented by bridges in a loop. They will
- also be connected at two points to routers on the backbone. The routers are
- routing IP, Decnet, and AppleTalk. They are also bridging LAT and MOP.
-
- Now, if the routers and the bridges use the same spanning tree protocol, then
- from the bridges point of view, the whole LAN is seen, and should a bridge or
- router go down in one part of the network then potentially it could have an
- impact elsewhere on another subnet. Actually, trying to think this through as
- I write this, it seems it shouldn't, but we were getting ELMS messages about
- topology changes until last night, when we made the LAN150's we have use the
- IEEE Spanning Tree. Those messages have stopped now, and we couldn't figure
- out what was causing them. It was as if each Bridge said, "I heard a rumor that
- there was a topology change". Only the routers are using the DEC Spanning Tree
- now, whereas before we made the change both the routers and the bridges were.
-
- We think this approach will give us greater flexibility, but I would like to
- hear from anyone who thinks this is a bad idea.
-
- We also have NAT bridges; they have a proprietary Spanning Tree, but are
- planning a new firmware version which will support DEC and IEEE. We are happy
- with their existing algorithm, which has one advantage as we see it: if a
- topology change occurs because power goes off in some building (say), and a
- backup bridge takes over, then once the power comes back the bridges do *not*
- automatically switch back. This means we can schedule when the switch back
- occurs. This is important for our important subnets, which typically have
- large VAX clusters on them, which run continuously. They don't like running
- with a long cluster-timout, so there is the potential for a cluster break if a
- bridge goes down. One such they don't like; to have two looks like
- carelessness.
-
- Tim Streater / SLAC Network Development.
-