home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.protocols.nfs:3199 comp.dcom.isdn:1213
- Path: sparky!uunet!olivea!sgigate!sgi!rhyolite!vjs
- From: vjs@rhyolite.wpd.sgi.com (Vernon Schryver)
- Newsgroups: comp.protocols.nfs,comp.dcom.isdn
- Subject: Re: Low cost ether/isdn brouters (was PC-NFS PPP Serial/ISDN driver wanted)
- Message-ID: <v74ftus@rhyolite.wpd.sgi.com>
- Date: 22 Jan 93 20:47:54 GMT
- References: <5da984b1.1bc5b@pisa.citi.umich.edu> <1993Jan21.224740.4259@gandalf.ca>
- Organization: Silicon Graphics, Inc. Mountain View, CA
- Lines: 61
-
- In article <1993Jan21.224740.4259@gandalf.ca>, dcarr@gandalf.ca (Dave Carr) writes:
- > ...
- > Try a 5220. And if you want a router with the same feature (data
- > compression, SNMP,...) for under $3000, wait a couple of months.
- > ...
-
- > Try the 5220i. Not a misprint. The 5200i can dial on demand, and only
- > brings the link up when needed. Shoot, it even looks at ARPs and IP
- > addresses. So tell me, is a router or a bridge.
- > ...
-
- > The 5220i can dial different ISDN numbers to get to different IP
- > addresses. Okay, it's a static router. Some limitations apply.
- > What's the equivalent CISCO box sell for? What's that, you want
- > compression. Sorry. We're Cisco, you don't need compression.
- > ...
-
- I mentioned cisco only to show what happened to the old style bridge
- makers of years ago, who insisted on making dumb bridges and only
- bridging. Remember what the notion of "brouter" did to them?
-
- $3000 would be expensive for a router. Late last year you could buy a
- T1 router (without CSU/DSU, but T1 and not measely 56K or 64K) for
- about half of that. Not (as far as I know) from Cisco, but then Cisco
- is a big, mature company now.
-
-
- > ..
- > >Finally, I notice that you posted this to comp.protocols.nfs. That
- > >suggests you are thinking about running NFS over your ISDN boxes. NFS
- > >usually implies NIS. NIS almost always involves broadcast packets.
- > >How are you going to "eliminate these [broadcast] packets"?
- >
- > Sure, why not. After all, a bridge doesn't discriminate on the basis
- > of protocol :-)
- >
- > From what angle? If bandwidth on the link is the problem, then I'll
- > ask "How much information content is in a broadcast? In the second?"
- > Know what compression does to these packets?
- >
- > If your assuming that I am going to filter all broadcast and therefore
- > NIS packets won't get through, that's a wrong assumption.
-
- I'm not sure what you're saying, but I think you should get some packet
- traces of big networks. There is much redundancy and so hope for
- compression as I infer that you hope. Unless the bridge is congesting
- things, the broadcasts are not simply retransmissions of the same stuff
- and so will not be compressed, at least not if your compression
- dictionary (or whatever) adapts to traffic faster than every 30 sec.
- (~2Mbit at 64Kbit/sec).
-
- The corporate network I know tends to have a lot of broadcast and
- multicast packets. Besides ARP and NIS, there are games (~30 p/s/player)
- and audio multicast streams. No plausible compression will be enough.
- You will have to filter by looking at higher layer protocols.
-
- Still and all, products like your 5220* sound like efforts in the
- right direction, so I'm not sure what we are arguing about.
-
-
- Vernon Schryver, vjs@sgi.com
-