home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.cell-relay
- Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!cs.utexas.edu!sun-barr!ames!kronos.arc.nasa.gov!iscnvx!news
- From: myoung@force.ssd.lmsc.lockheed.com
- Subject: Variable Length Issues
- Message-ID: <1992Nov19.153830.28505@iscnvx.lmsc.lockheed.com>
- Sender: news@iscnvx.lmsc.lockheed.com (News)
- Reply-To: myoung@force.ssd.lmsc.lockheed.com
- Organization: LMSC, Sunnyvale, California
- Date: Thu, 19 Nov 92 15:38:30 GMT
- Lines: 66
-
- In article <22909@venera.isi.edu>, finn@isi.edu (Greg Finn) writes:
- >
- > Very true. I co-developed a multi-gigabit LAN. If a
- >store-and-forward (or retrieve) operation is to occur without reducing
- >the channel bandwidth, it requires memory that operates at twice the
- >channel rate.
-
- I agree, this a valid complaint about ATM. However see below the
- argument on cut through switching.
-
- But, having the capability for store and forward allows switches to
- be connected in random mesh patterns without having to build the
- cut through operation through an arbitrary number of nodes.
-
- >
- > Packet storage within a switch should be avoided at high
- >channel rates if practical and can be accomplished with cut-through
- >routing. This issue has nothing whatsoever to do with fixed versus
- >variable-length packets.
-
- Cut through is a little related to the fixed vs variable issue: cell
- relay like cut-through switching in small chunks. But in essence I
- agree, having small fixed sizes does not prevent cut through switching,
- nor does it prevent all optical designs.
-
- >I can say something about LANs. PTM allows one to design LANs that are
- >exceedingly fast and that have tiny host interfaces that
- >require no fragmentation/reassembly hardware.
-
- And this is the strongest criticism, but:
-
- Fragmentation allows slow speed devices to share the same switching
- bandwidth with higher speed devices, without bridging and allows arbitrary
- mesh networks. This is valid even in single sight installations.
-
- Variable length with cut-through requires all devices to have channels
- at the same high rate to prevent congestion. How do you design the same
- LAN to serve low cost PCs or high performance sstations without
- wasting the technical gains by adding bridges. Bridges are now
- quite common, and they include all the management hassles.
-
- > ATM is politically very popular. Arguments that ATM brings
- >about end-to-end compatibility do not impress me. Most LAN traffic
- >stays within the LAN.
-
- Ah, but the most valuable (monetary) user traffic is often
- the WAN traffic, like distributed sales and inventory.
-
- **************
-
- Finally, what about the argument for dual priority and support for
- short low latency messages?
-
- **************
- Additional Comments:
-
- Variable length cut through certainly has a place in a perfect world,
- unfortunately its not a perfect world and the two technologies will
- compete, and cell relay will compete effectively.
-
- Perhaps we need a standard which allows a cell relay virtual path to
- reserve enough synchronous bandwidth to feed a variable length LAN
- on the run, performing the re-assembly on the fly without a message
- store and forward?
-
- And dare we raise the cell size issue again, egad!
-