Cell Relay Archive

Cell Relay Retreat>List Archive>month:1999-Jun> msg00069



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: ILMI address registration question

  • From: albert.e.manfredi@boeing.com
  • Date: Sun, 13 Jun 1999 23:54:05 GMT
  • Organization: Boeing North American
  • X-Article-Creation-Date: Sun Jun 13 23:54:05 1999 GMT


In article <376131E2.75E70398@xedia.com>,
  Paul Koning <pkoning@xedia.com> wrote:
> Thomas Betker wrote:
> >
> > Shaheen Ali (ali@mediaone.net) wrote:
> > : I have a general ILMI address registration question.  If a host
on the
> > : user side of a UNI connection (DTE) has multiple addresses to
register,
> > : do all the addresses have to have the same ESI?
> >
> > : So, for example.  If I have two services on my host (one LANE
server and
> > : one IPOA client for example):
> >
> > : ESI                        SEL
> > : 00.08.11.00.00.01.01
> > : 0A.08.11.00.00.01.02
> >
> > : Notice that the ESI fields are different.  If after getting a
network
> > : prefix, may I register the above two addresses with the upstream
> > : switch?  Will the switch be able to deal with the fact that one
DTE has
> > : two addresses with different ESI fields?
> >
> > According to the ILMI Spec., you may register more than one ESI.
> > This may be restricted by the switch's implementation, though.
>
> Also, I would question why the end system is using two different ESIs.
> The example shown is technically legal since the second ESI has the
> "locally administered" flag set.  But it seems a *very* questionable
> practice.
>
> The whole purpose of the SEL byte is to allow multiple addressable
> entities to exist at a given host without consuming multiple ESIs.

Hmm, Paul. I don't know that you can assume that the "local/global"
flag used in MAC addresses also applies to ESIs? Not that this changes
anything here, but the ESI is really not constrained to any MAC address
rules, as far as I can determine. Perhaps I missed some significant
paragraph somewhere.

My reaction in general would be that what Shaheen wants to do is
perfectly fine. Although I also agree that the SEL's purpose is as you
describe. (But should you need more than 256 internal destinations for
these messages, you might need to resort to using the ESI bits.)

--
Bert
manfredi@arl.bna.boeing.com


Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.