home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.cell-relay
- Path: sparky!uunet!mcsun!dxcern!brian
- From: brian@dxcern.cern.ch (Brian Carpenter CERN-CN)
- Subject: Re: Future of IP routers
- Message-ID: <1992Aug31.133020.10227@dxcern.cern.ch>
- Organization: CERN European Lab for Particle Physics
- References: <3445@ra.nrl.navy.mil> <1992Aug27.174448.28143@fokus.gmd.de> <1992Aug28.092320.17105@ccsun.strath.ac.uk> <9224411.15484@mulga.cs.mu.OZ.AU> <1992Aug31.091218.21819@fokus.gmd.de>
- Date: Mon, 31 Aug 1992 13:30:20 GMT
- Lines: 50
-
- In <1992Aug31.091218.21819@fokus.gmd.de> icarus!mbu@fokus.gmd.de (M. Burak) writes:
-
- Markus:
-
- >Brian Carpenter writes:
-
- >>Sorry, as usual CCITT and their friends are too far behind reality.
- >>IETF and the ATM Forum will have resolved all these questions before
- >>CCITT even formally votes on AAL5.
-
- >Sounds very good. But please tell me how will IP work over an ATM network with
- >AAL type 5? What is the concept of this approach?
- >Is connection setup required prior to sending CL-data or do you assume a fully
- >meshed network? What function in the ATM network will perform routing of IP packets if you
- >only have ATM switches providing an ATM layer?
- >I have no idea. Is there anybody who can help me to answer these questions?
-
- The IETF has to develop answers to these questions, and this will
- most likely take place in a bottom-up process. My guess is "yes", i.e.
- routers, fully-meshed subnets and dynamic call setup will coexist.
- In the dynamic case each host has to contain call setup logic
- and a timer to tear down calls that have been idle for too long.
-
- The work done in IETF so far is on the lowest level, how IP packets
- will be mapped onto AAL5 PDUs. Once this draft is stable it could
- be used immediately in pre-configured networks between hosts and/or
- routers - i.e. in any context where the VCI/VPI to reach the next
- system is known. (It would also work for CLNP, IPX etc as far as
- I know.)
-
- The next stage is "for further study" and I suspect that not much
- will happen until Q.93B is stable. I have personally developed the
- view that the next stage in the bottom-up design process is to enhance
- DNS to be capable of returning arbitrary switched network addresses
- as well as IP addresses: initially these could be VCI/VPI values,
- later on E.164 or whatever. Then the procedure to establish an IP
- based session to foo@bar.sna.fu would be
- - issue DNS request for that address
- - if reply indicates a specific VCI/VPI pair, use it immediately
- - if reply indicates an E.164 address/subaddress, invoke Q.93B
- to establish VC/VP (unless already open)
- - else go to non-ATM processing
-
- You will notice this does not need ARP (see my previous posting)
-
- Comments on this are wanted!
-
- Regards,
- Brian Carpenter CERN, brian@dxcern.cern.ch
- voice +41 22 767 4967, fax +41 22 767 7155
-