home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.cell-relay
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!sdd.hp.com!spool.mu.edu!agate!boulder!ucsu!tramp.Colorado.EDU!barr
- From: barr@tramp.Colorado.EDU (BARR DOUG)
- Subject: Re: LAN link cost issues
- Message-ID: <1992Dec22.185414.25469@ucsu.Colorado.EDU>
- Sender: news@ucsu.Colorado.EDU (USENET News System)
- Nntp-Posting-Host: tramp.colorado.edu
- Organization: University of Colorado, Boulder
- References: <to6efqs@sgi.sgi.com> <23095@venera.isi.edu>
- Date: Tue, 22 Dec 1992 18:54:14 GMT
- Lines: 48
-
- In article <23095@venera.isi.edu> finn@dalek.isi.edu (Greg Finn) writes:
- >In article <to6efqs@sgi.sgi.com> rpw3@rigden.wpd.sgi.com (Rob Warnock) writes:
- >
- >>30 meters is too short. The generally-accepted "desk to wiring-closet"
- >>requirement is 100 meters, and even that is marginal.
- >
- > Not really. 100 meters is an encompassing requirement. 30
- >meters gets you a sizable percentage of point-to-point link distances
- >in practice. In our office tower it would handle most workstation to
-
- Maybe. But have you ever had to choose a system? When choosing a system
- we are very unwilling to choose one that will not offer services to ALL
- our clients. What do you say to that one person down the hall too far?
- His manager will say "Well, it's a good system except for the fact that
- they can't service one of our critical people." There is also a STANDARD,
- 100 meters, EIA/TIA 568, and if I can't provide service because of a loong
- run I can always cover my ass by saying "Well, it's out of spec, EIA/TIA
- 568 says 100 meters or less." And conversly the client can say "What, you
- can't service wire plants that meet EIA/TIA 568? See ya."
-
- >concentrator links now. If for a good percentage the interface cost
- >is several hundreds of dollars less ... you would choose that, no?
- >There remains the ability to site additional concentrators also. The
- >desirability would depend upon the cost savings. Four sites per floor
- >is sufficient for complete geographic spread in a good sized
- >multi-hall office building.
- >
- > This is all speculatation though. When we have proven the
- >cable technology I can state with reasonable reliability the interface
- >and chip costs. I should know in about three to four months.
- >
- >>Yes, LLNL is very vocal about their approach. Yet when you talk to other
- >>HIPPI users besides LLNL, you get a different view. (The "Xunet" testbed,
- >>for example, is using an ATM-based fabric.) I'm not sure how far one vocal
- >>user can drive the market...
- >
- > The true gigabit LAN arena is a small one at present. LANL is
- >more a HIPPI proponent. Work is proceeding on HIPPI LAN standards.
- >None of these have inexpensive interfaces yet. Fibre Channel
- >addresses that via much slower alternative channel rates, but using
- >that bumps you out of the gigabit arena.
- >
- >--- ggf
- >--
- >Gregory Finn (310) 822-1511
- >Information Sciences Institute, Marina Del Rey, CA 90292
-
-
-