Hi all! First, I would like to thank all the people who tracked my thread about DNS SRV :) I scanned the SIP draft looking for something about when to do and when not to do DNS resolution (for resolving various SIP addresses) upon message transmission to the next-hop. IMO this point is not clear... no strict rules seem to be expressed in the draft (I tried to read it carefully :) Of course when I say "not doing DNS resolution" I mean using previously resolved name and not doing request to gather an update (if it exists).. Yet besides any name-resolution capabilities, DNS may be useful/required to support DNS-based load-sharing features (like we do with www) and anyway name-resolution/translation is one of the most important bricks of the SIP framework.. So do we do requests to DNS for each outgoing message, for each transaction..etc etc? and BTW, DNS is not available on all targeted platforms.. That's why I've been thinking that DNS usage was implementation dependent, since the draft does not show any strict protocol rules; and so SIP implementations MUST NOT state that one other is doing/or not doing DNS request at a given time... What do you think about all of this? Is there any other threads about load-sharing or/and high-availability of SIP nodes? I did not see a lot of people speaking about the SIP protocol design concerning these 2 points (they are not much emphasized) Thanks --Frederic ROLLAND COMPAQ Telecom Engineering Dept. Sophia-Antipolis (FRANCE) _______________________________________________ Sip-implementors mailing list [EMAIL PROTECTED] http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors
