Taisto,

I think you need to define what you mean by redundancy. If you want to
be able to continue a call you would probably have to do DNS NAPTR (or
SRV) lookups for every routing decision. However, the biggest problem
here would be maintaining the call state over the proxies/b2bua that are
in that DNS list. I guess the replication traffic would be terrible.

I would assume that it is enough if you maintain the registration state
of the user agents across the server farm. The server then can insert
its IP address and then all call messages will flow over this specific
proxy/b2bua. When this instance fails, users will look into their
handset and make another call (which should be acceptable if the uptime
of a server is not too bad).

For maintenance, you can take a server out of the DNS list first and
then there will be no calls that are interrupted.

Best,


Christian
--
Dr.-Ing. Christian Stredicke
sip:[EMAIL PROTECTED]

> -----Urspr�ngliche Nachricht-----
> Von: [EMAIL PROTECTED] [mailto:sip-implementors-
> [EMAIL PROTECTED]] Im Auftrag von Taisto Qvist
> Gesendet: Mittwoch, 12. Februar 2003 21:11
> An: [EMAIL PROTECTED]
> Betreff: [Sip-implementors] Clarification on Proxy or B2BUA and
> DNS/Via:/NAPTR
> 
> Hi SIP Experts!
> 
> I was wondering wether anyone could help me clarify the specific
> requirements for a B2BUA.
> I know it is supposed to act just like two UA's back to back, just
> like the name says, but when stumbling on the notes regarding 3GPP's
Rel
> 5
> and its CSCF's I find my self slight fuzzy on the differences.
> 
> These CSCFs are "SIP-incompliant" because they send messages on
account
> of the endpoint, behaving as if the _were_ that specific endpoint.
(they
> can send BYE i think one of the examples were)
> Now I understand why they shouldnt, if we talk S/MIME security and so
> on,
> but I cant exactly understand the finer details of what the proxy
would
> do
> differently if it did implement/use all the requirements of a B2BUA.
> 
> My other question is a simple one regarding via-headers and their use.
> One of the benefits of having the via-header route the response
> "backwards",
> towards the originator, is that of redundancy the rfc says. (apart
from
> not needing
> state for remembering where it came from)
> So if the host in the via-example below doesnt answer, the proxy could
> try
> doing more DNS lookups, trying to find an alternative right?
> My simple question is...should this be done on the full name
> "server42.betelgeuse.com", or just the domain name "betelgeuse.com" ?
> Should each server have extra NAPTR records for all the possible
> backupservers?
> What about the recieved=x.x.x.x line?
> 
> And then I final check..this redundancy can only work properly in
> non-stateful
> environments right? Since the backup proxy wouldnt have neither dialog
> nor transactional
> state, unless conveyed by outofbounds means?
> 
> Via: SIP/2.0/UDP server42.betelgeuse.com
>          ;branch=z9hG4bKnasads8;received=192.0.1.2
> 
> 
> Very thankful for any and all helpful clarifications
> 
> Regards
> Taisto Qvist
> _______________________________________________
> Sip-implementors mailing list
> [EMAIL PROTECTED]
> http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors


_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to