RE: draft-ietf-mmusic-sip-session-timer-00.txt

[Douglas Clowes <[e-mail]>]

At 07:40 1999-02-16 -0700, Donovan, Steven R. (MCI) wrote: 

>>>>

<excerpt>

<fontfamily><param>Fixedsys</param><color><param>0000,0000,ffff</param><smaller>...<<snip></smaller></color></fontfamily> 


<smaller>IRR is being used in H.323 to determine if the endpoint is still
alive, but</smaller> 

<smaller>the preference seems to be heading toward gatekeeper routed
calls. Here,</smaller> 

<smaller>you have a TCP connection through the gatekeeper for session
control, but</smaller> 

<smaller>media still goes direct.</smaller> 


<fontfamily><param>Fixedsys</param><color><param>0000,0000,ffff</param><smaller>Doesn't
this have some real scaling and performance problems?  I assume that this
requires setting up a TCP connection per call attempt and limits the
number of sessions that a single gatekeeper can handle to the number of
simultaneous TCP connections that the gatekeeper can handle.

</smaller></color></fontfamily>

</excerpt><<<<<<<<

Well, yes, if you want to state the obvious. Actually, from memory it's
two (or more - I seldom get down to that level) but it begs the question
of the role and scope of a gatekeeper. If you have one gatekeeper per
eight line gateway, it's not a problem. If a gatekeeper's role is keeping
the gate, and its scope is a single gateway or POP up to a few thousand
lines, it's probably still not a problem.


If, OTOH, you want it to be a network controller for a major telco ...
different story.


Regards,


Douglas

>>>>

<excerpt>

<smaller>>-=Francois=-</smaller> 

<smaller>>[e-mail]</smaller> 

<smaller>></smaller> 


<smaller>Douglas</smaller> 


</excerpt><<<<<<<<