home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!think.com!sdd.hp.com!usc!isi.edu!finn
- From: finn@isi.edu (Greg Finn)
- Newsgroups: comp.dcom.cell-relay
- Subject: Re: Computers dont like ATM?
- Message-ID: <23067@venera.isi.edu>
- Date: 16 Dec 92 05:57:26 GMT
- References: <tl6jc74@sgi.sgi.com>
- Sender: news@isi.edu
- Reply-To: finn@dalek.isi.edu (Greg Finn)
- Organization: USC-Information Sciences Institute
- Lines: 58
-
- In article <tl6jc74@sgi.sgi.com> rpw3@rigden.wpd.sgi.com (Rob Warnock) writes:
-
- >Currently the big holdup is the ODL prices, but when those plummet ("soon"),
- >the prices for the gigabit encoder/decoder chips will be the bottleneck.
- >*That's* what needs work (e.g., price reduction).
-
- Ahh. I see now your misunderstanding or my lack of clarity.
- I worry about building gigabit NETWORK links. That includes the
- framing, encoder/decoder, and for 1300nm fiber, the OFC logic needed
- to meet Class I compliance. A typical gigabit fiber link for a
- network needs a framer, an encoder, a decoder, an optical transmitter,
- a detector and an OFC circuit. If you are point-to-point you need two
- sets to be duplex.
-
- I can't get there from here today unless I spend circa $1500
- for parts alone. But I can get the parts. What you call the ODL is
- not the majority of the cost. I agree with you that gigabit
- encoder/decoder chips are costly.
-
- >In fact, a lot of the enthusiasm for ATM in LANs sprang originally from
- >the hope that with the telcos using ATM, the componenets would be in mcuh
- >higher volume that mere LAN use could generate, and thus would be cheaper.
- >Ironically, it is the LAN applications that are now pushing the deployment
- >schedule for ATM, so the components won't see volume production by the time
- >the first ATM LANs start being installed... so prices will be high (at first).
-
- I do not see 655 Mb/s OC-12 being pushed commercially for LANs
- for quite some time. Some major computer manufacturers are moving
- toward 155 Mb/s and decommitting from their originally rosy prediction
- of OC-12. I guessed publicly that it was because OC-12 rates require
- costly gigabit links ... meaning gigabit network links. I still think
- that it is a good guess.
-
- 155 Mb/s with ATM's encoding overhead gets me somewhere in the
- neighborhood of 80 Mb/s of TCP/IP payload in today's Internet. Hardly
- exciting. It won't get you the headroom that you need for studio
- NTSC.
-
- For bandwidth in the gigabit domain for LANs, it looks to me
- as if Fibre Channel may have a commercial leg-up on ATM in the gigabit
- LAN arena. My problem with Fibre Channel is that it makes the
- overhead of ATM carrying Internet packets look quite small by
- comparison.
-
- There is no shortage of players talking a gigabit LAN game. I
- see a shortage of delivery however. That is true of my effort too.
-
- We are short on longer distance cabling technology, but have
- inexpensive gigabit LAN switching and interface technology running
- now. Silicon development on cabling/net-maintenance is proceeding.
- ATM has 155 Mb/s LAN service available commercially, but gigabit
- service is a somewhat distant promise. Fibre Channel will shortly
- deliver on gigabit link speed, but is very weak on switching.
-
- It will be an interesting next two years.
- --
- Gregory Finn (310) 822-1511
- Information Sciences Institute, Marina Del Rey, CA 90292
-