Cell Relay Archive

[Victor Shvetsky <victor@ablecom.co.jp>]

I might be wrong, but 1 second delay is obtainable via change of timeout
timer for the server/bus/client in LANE, for example. 35 sec is
definitely unacceptable. The trick is whether or not you are using PNNI
or BICI. Remember that in PNNI almost everything can be reconfigured
based upon what you like.

Hope this answered some of your questions,
Vic

dingus@ibm.net wrote:
> 
> In <847378637.12460.0@martegdm.demon.co.uk>, atwell@martegdm.demon.co.uk writes:
> >Does anyone know of any figures for the time to reroute over ATM.  We
> >have tried this and it has taken around 35 seconds which is no good
> >for the real-time applications we are intersted in.
> >The setup we have is two ATM switches and a workstation at either end.
> >We connected two backbone cables between the ATM switches and timed
> >how long it took to reroute when one was disconnected.  We are looking
> >for a figure of around 1 second.
> >Is this realistic?  If not how is it possible to have redundancy over
> >ATM for real-time applications?
> >
> What protocols? AAL? etc. 35 seconds sounds about right for a total
> resetup of LANE stuff (maybe a bit on the high side).
> 
> 1 second is probably hard with current SVC timeouts and reconfiguration
> delays. I think I have seen approx a 10 second reconfiguration...
> 
> Francis Turner
> Madge US - All opinions are my very own though
> fturner@madge.com