home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.cell-relay
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!wupost!usc!rpi!batcomputer!munnari.oz.au!cs.mu.OZ.AU!mullian!gja
- From: gja@mullian.ee.mu.oz.au (Grenville Armitage)
- Subject: Re: Future of IP routers
- Message-ID: <9224411.15484@mulga.cs.mu.OZ.AU>
- Sender: news@cs.mu.OZ.AU
- Organization: Basket Weavers, Inc.
- References: <3445@ra.nrl.navy.mil> <1992Aug27.174448.28143@fokus.gmd.de> <1992Aug28.092320.17105@ccsun.strath.ac.uk>
- Date: Mon, 31 Aug 1992 01:33:41 GMT
- Lines: 53
-
- In article <1992Aug28.092320.17105@ccsun.strath.ac.uk> craa85@ccsun.strath.ac.uk ( D.W.Stevenson) writes:
-
- I haven't seen a response to this, so I'll post...
-
- >The point I am trying to make is that IP is
- >designed to route packets, and by route I mean
- >direct the packet to the next router on it's
- >journey. Functionality of the ATM layer seems to
- >perform the same function (substitute 'ATM' for 'IP',
- >'cells' for `packet` and 'switch' for 'router'
- >in the last sentence)
- >and hence there is a wasteful overlap in
- >functionality when you have to reassemble
- >at each intervening router for router software to process.
-
- Well, there is and there isn't (a wasteful overlap). IP was designed
- to route primarily between separate (and/or different) physical layer
- technologies. For various reasons routing gets done on a per-packet
- basis, etc, etc. However ATM should, in reality, be seen as primarily
- another physical layer packet transport mechanism. Routing in the
- ATM sense does not actually occur on a cell by cell basis - the
- route that a virtual circuit follows is established at the beginning
- of the call. A much more basic function of cell _switching_ occurs
- during a virtual circuits operation, but this is not at the same level as
- IP style routing.
-
- I think a distinction needs to be made between 'routing' when finding
- a route, and 'routing' (or switching) when retransmitting a data unit
- (packet or cell) out onto the physical media chosen as the correct route.
-
- [..]
- >Surely once I can open an ATM connection
- >from my workstation to your workstation,
- >passing across the world, through the global ATM network of switches,
- >we then become connected to the same network
- >and so no internet layer is needed and IP routers or equivalent
- >will no longer be required for data transfer over the ATM network.
-
- In the same way that you don't need an IP router between workstations
- on the same Ethernet, you are right. However the world will never be fully
- ATM. Even ATM-like LANs may not be able to switch cells seamlessly with the
- telcos B-ISDN offerings, necessitating some form of gatewaying. Two places
- where IP routing functions _will_ be needed are:
-
- - At the edges of the ATM networks (i.e. between ATM and some LAN, etc)
- - Between virtual subnetworks within the ATM world, which may exist for
- varieties of political, security, or financial reasons.
-
- ATM, meaning AAL1 through AAL5 (and not including I.364, CCITTs
- connectionless service), is layered underneath protocols such as IP.
- It is not positioned to replace them.
-
- gja
-