[prev in list] [next in list] [prev in thread] [next in thread] 

List:       voipsec
Subject:    Re: [VOIPSEC] TLS as the SIP security mechanism
From:       Robert Moskowitz <rgm () icsalabs ! com>
Date:       2005-08-19 21:07:42
Message-ID: 6.2.3.4.2.20050819165315.03371e58 () localhost
[Download RAW message or body]

At 07:43 PM 8/15/2005, Pankaj Shroff wrote:
>What does the history of DNS outages tell us about SIP reliability? Other
>than the fact that DNS is a service at the IP core (with ENUM likely to be
>integrated in a similar pattern), all that says is that DNS is an essential
>service which should seldom fail. From a Voice-over-IP point of view, the
>comparison is more appropriate to the local name resolution servers that
>fail a lot more often. VoIP is not (considered to be) a core network
>service.

I have been following this thread, and every now and then thought 
perhaps I should put my 2 plug-nickels into the fray...

Today is NOT 1994.  Gee I remember what Mosaic did to the servers 
when it started running 4 concurrent TCP connections.  A few of us 
had to teach Andresen queueing theory to get him to understand what 
he did.  But the server vendors finally figure out how to support 
more than 128 concurrent TCP connections.

There are many good examples today of what is needed for a robust 
application infrastructure that SIP requires.  Today there are blades 
and SANS that make very dense, very distributed, very robust 
configurations possible and probably affordable.

In '87 I was at a breakfast where Andy Grove explained that he no 
longer concerned himself, for planning purposes, where engineering 
would get the next idea to maintain Moore's Law.

Don't sweat the details.  Design what meets the goals.

What is needed from here is a clear, concise review of various levels 
of outages will have on SIP services.

Brian alluded to a few of them.

It would be good to have senarios like:  neighborhood power outage 
for 8 hours, where the CO stays up and whole blocks at a time get 
power back and everyone hits their phones at once.

This is a very common event here in Michigan, and I would suspect 
that it is common throughout the US.  CPE batteries might not last 8 
hours, BTW.

These, I would hope, go into what a provider will be putting into their plans.


Robert Moskowitz
Senior Technical Director
ICSA Labs, a division of Cybertrust, Inc.
W:      248-968-9809
F:      248-968-2824
VoIP:   248-291-0713
E:      rgm@icsalabs.com

There's no limit to what can be accomplished if it doesn't matter who 
gets the credit



_______________________________________________
Voipsec mailing list
Voipsec@voipsa.org
http://voipsa.org/mailman/listinfo/voipsec_voipsa.org
[prev in list] [next in list] [prev in thread] [next in thread]