home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!elroy.jpl.nasa.gov!usc!isi.edu!finn
- From: finn@isi.edu (Greg Finn)
- Newsgroups: comp.dcom.cell-relay
- Subject: Re: Computers dont like ATM?
- Message-ID: <23082@venera.isi.edu>
- Date: 17 Dec 92 20:02:55 GMT
- References: <tmp0lkg@sgi.sgi.com>
- Sender: news@isi.edu
- Reply-To: finn@dalek.isi.edu (Greg Finn)
- Organization: USC-Information Sciences Institute
- Lines: 54
-
-
- In article <tmp0lkg@sgi.sgi.com> rpw3@rigden.wpd.sgi.com (Rob Warnock) writes:
- >finn@dalek.isi.edu (Greg Finn) writes:
-
- >Well, until the "new" ODLs ("Optical Data Link", a transmitter/receiver pair)
- >come out, the best I have seen for a gigabit is *$3000* the pair, *plus* your
- >$1500 for other chips. (The new ODLs will drop the ~$3000 to $500 or less...)
-
- So your observed prices are the same as mine essentially. You
- think that the ODL chips will drop to $500/pair + the encoder/decoder.
- That will not reduce the cost of a gigabit fiber network link to
- reasonable levels for workstations.
-
- For short runs of byte wide twisted pair, some of the new
- differential drivers can produce at least 500 Mb/s payload for about
- $50/pair in handfull quantity. Point-to-point runs of up to 30 meters
- buys you most of the office to wiring closet connections. We are
- working in this area now.
-
- >Well, the HIPPI market is looking quite eagerly at 622 Mb/s ATM (or so
- >we've been told), because HIPPI fiber-optic extenders are so expensive
- >(over $10K per end). [HIPPI only goes 25m in copper.] And most HIPPI
- >users would be willing to settle for 622 instead of 800, given that
- >few systems can drive at the full 800 Mb/s anyway...
-
- The HIPPI market seems to be implementing HIPPI on Fibre Net.
- LLNL is moving that way. The vendors are IBM, Honeywell, Psi Tech and
- Ancor. Fibre Channel is specifically designed to handle the HIPPI
- payload as its upper-limit option, hence its attractiveness.
-
- >Well, as somebody pointed out to me, with SONET coding the actual cell
- >payload data rate is 135.63 Mb/s, substantially *above* FDDI, especially
- >when you consider it's full-duplex continuously.
-
- IP into 53-bytes produces a roughly 50% penalty with the
- existing Internet traffic mix. That has been discussed here before.
- FDDI is less attractive than ATM for other reasons that encoding
- overhead I agree.
-
- >+---------------
- >| It won't get you the headroom that you need for studio NTSC.
- >+---------------
-
- >The people I know who do that sort of stuff think so, with within-frame
- >JPEG compression.
-
- Maybe. I have heard that now for some years. I am told from
- conferences that the digital studio efforts are not using JPEG. The
- MPEG experience has not been very good so far. I could be a
- spoil-sport and point out that with a gigabit LAN and a multi-Gb/s
- switch you don't need video compression.
- --
- Gregory Finn (310) 822-1511
- Information Sciences Institute, Marina Del Rey, CA 90292
-