Francois,
Can I explain why you believe that for "small" servers (supporting < 20
clients) you believe
that the retry after is problematic? I don't think it's that clear from
the draft.
The problem of 503 with Retry-After is described in greater detail in
Sections 4.1 and 4.3 of draft-ietf-sipping-overload-reqs-00. In short,
503 with Retry-After may lead to the oscillation of traffic a server
receives and the shifting of load between servers.
And if so, would there be a way to mitigate the oscillation while
allowing the Retry-After?
The proposed changes 5. and 6. would still allow a Retry-After header,
although it would be restricted to a single request.
An alternative to proposed change 1. (i.e. to not recommend the use of
Retry-After in 503) could be use a Retry-After value based on
milliseconds. This would provide a more fine grained control to
overloaded proxies and may avoid some of the problems the current header
has. It would work for servers irrespective of the number of clients.
However, it would change the semantics of the Retry-After value.
Overall, the goal of this draft is not to introduce a new overload
control mechanism (which certainly should avoid traffic oscillations)
but rather to see how far we can get with modest changes to the existing
spec.
I'm a little concerned about a rule that seems somewhat vague on this.
-----Original Message-----
From: Volker Hilt [mailto:[EMAIL PROTECTED]
Sent: Thursday, May 17, 2007 10:11
To: sip
Subject: [Sip] [Fwd: I-D ACTION:draft-hilt-sip-correction-503-00.txt]
Below is a draft that proposes a correction to the definition
of 503 (Service Unavailable) responses in RFC 3261. The goal
of this draft is to avoid some of the problems that have been
identified with 503, however, it does not attempt to provide
a full blown overload control solution.
http://www.ietf.org/internet-drafts/draft-hilt-sip-correction-
503-00.txt
Comments, thoughts are highly appreciated.
Thanks,
Volker
_______________________________________________
Sip mailing list https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip
--
Volker Hilt Bell Labs / Alcatel-Lucent
phone: +1 732 332 6432 101 Crawfords Corner Rd
email: [EMAIL PROTECTED] Holmdel, NJ 07733
_______________________________________________
Sip mailing list https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip
