[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [Sip] Impossible to use real persistent connection with sip



I look forward to reading the draft. I have been hoping someone would
write some advice to implementers on when to close connection. 

I initially did not understand that pc was meant to be a hint - I'm way
more comfortable with a hint than with anything that assumes a
reliability model beyond what SIP already has specified.

Cullen
 

> -----Original Message-----
> From: sip-admin@ietf.org [mailto:sip-admin@ietf.org] On 
> Behalf Of Vijay K. Gurbani
> Sent: Monday, February 10, 2003 8:21 AM
> To: Cullen Jennings
> Cc: Colasanto, Eric (Eric); 'James Ford'; Jain, Rajnish 
> (Rajnish); sip@ietf.org; Sarit Galanos Mekler
> Subject: Re: [Sip] Impossible to use real persistent 
> connection with sip
> 
> 
> Cullen Jennings wrote:
> > Just to get a little more precise about this ... what does 
> persistence
> > mean?  For the length of the transaction? For the dialog? As long as
> > the registration is valid? Until some box crashes? Until 
> the world moves 
> > to IPv6.
> 
> Hi Cullen:
> 
> Yes, we've grappled with that as well.  Clearly, to get most 
> bang for the buck, the persistent connection lives beyond the 
> dialog that established it; i.e. it spans transactions.  One 
> place where such persistent connections may be used is 
> between a UAC and its default outbound proxy, or between two 
> proxies that may have a trusted peering relationship between 
> them for traffic between their domains.
> 
> Exactly how to quantify this length of time is tricky.  As 
> you point out, sometime a peer may go down for administrative 
> reasons, other times it may simply crash.  Depending on how 
> we specify and implement it, there may be an expiry time for 
> this connection -- I don't know what the correct answer is right now.
> 
> > I would be very happy to see some good thought on when 
> devices should
> > close a TCP socket - I think that figuring out a good algorithm and 
> > understanding the large scale consequences is going to be 
> surprisingly 
> > complicated.
> 
> We're working on an I-D that extends the connect-reuse ID to 
> outline some of these issues.  I will post it to the list as 
> soon as I can.  It contains a lot of what Sarit also outlines below:
> 
> Sarit Galanos Mekler (Sarit@radvision.com) writes:
> > First I would like to say that I think that the pc=1 is a 
> good choice. 
> > To my opinion adding the pc=1 to the via header is only a 
> > recommendation to the server to keep the connection open.
> 
> Unfortunately, this is just a hint to the server, and as you point out
> below:
> 
> > Servers that don't understand pc or don't want to keep the 
> connection
> > open are more then welcome to close it.
> 
> But that may be all we need to do.  I don't yet know if it is 
> worthwhile to go through some negotiation mechanism (i.e. 
> register Require and Proxy-Require tokens with IANA).  Even 
> it we do go the "pc=1" parameter route, we should register 
> the "pc" token with IANA.
> 
> > A server can also decide that it wishes to keep the connection open 
> > only
> > if it is actually being used.  An implementation can for 
> example, set a 
> > timer on server connections.   When ever a message is 
> received/sent on
> > the connection the timer is reset. when the timer expires 
> the server 
> > closes  the connection.
> 
> All this presumes that clients and servers are well-behaved.  
> In the extreme case, I think that we may rely on I/O failing 
> on a now unconnected socket because one of the peers crashed 
> or got brought down for adminstrative reasons.  That is, we 
> proscribe behavior for well- behaved peers, taking in account 
> extreme cases where one of them suddenly disappears.
> 
> > The persistency level (dialog / transaction) if for the specific
> > application to decide. For dialog persistency the 
> application will have 
> > to keep the connection open for the duration of the call.  
> Using the 
> > pc=1 parameter will give a good chance for using 
> connections for more 
> > then one transaction. An option that today (almost) does not exist.
> 
> Right.
> 
> Cheers,
> 
> - vijay
> -- 
> Vijay K. Gurbani  vkg@{lucent.com,research.bell-labs.com,acm.org}
> Wireless Networks Group/Internet Software and Services
> Lucent Technologies/Bell Labs Innovations, 2000 Lucent Lane, Rm 6G-440
> Naperville, Illinois 60566     Voice: +1 630 224 0216
> 
> _______________________________________________
> Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
> This list is for NEW development of the core SIP Protocol
> Use sip-implementors@cs.columbia.edu for questions on current 
> sip Use sipping@ietf.org for new developments on the 
> application of sip
> 

_______________________________________________
Sip mailing list  https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use sip-implementors@cs.columbia.edu for questions on current sip
Use sipping@ietf.org for new developments on the application of sip