The Routing Over Large Clouds Mailing List Archive by date

[fddi1-ncd@Baynetworks.COM (Andrew Smith)]

> Date: Mon, 29 Apr 1996 15:44:43 -0400
> To: rolc@nexen.com
> From: Derya Cansever <dhc2@gte.com>
> Subject: Limits on SVCCs

Derya,

> One of the considerations in using NHRP is to be selective in establishing
> short-cuts. This is motivated by at least two reasons: One, it is not worth
> to go through the signaling procedures when the application is of short
> duration. The other one is that, in all likelihood, there is a limit on the
> number of SVCC that an ATM host or an end user router with ATM interface can
> establish. Is there any estimates/references for this limit? Is this limit
> best expressed in terms of total SVCCs at a given time, or as a rate, such
> as SVCCs/sec?

I suggest talking to your vendors about their products, as Joel suggested.
Both the actual limit and the peak setup rate are important parameters. The
peak setup rate is especially important in terms of availability: how long
does it take to re-route or re-establish all the control and data calls
after a failure? How long does it take to reboot the network after power-failure? 
When shortcut VCCs are widely used, you'd better plan on long outage times
or else on expensive ATM switches. Using connectionless, multi-protocol, packet
relay points as a scaling mechanism for VCC usage then doesn't seem like that 
bad an idea ..... :-)

[ In fact a lot of the protocol design decisions taken for PNNI, 1577, NHRP,
MARS, MPOA and LANE (v1 and v2) have taken implementation reality into account. ]

> Derya

Andrew

********************************************************************************
Andrew Smith					TEL:	+1 408 764 1574
Bay Networks, Inc. 				FAX:	+1 408 988 5525
Santa Clara, CA					E-m:	asmith@baynetworks.com
********************************************************************************