[Summit] Mobile optimizations: presence queuing example

Joe Hildebrand joe.hildebrand at webex.com
Wed Feb 10 19:39:06 CST 2010


Aside from the whitespace ping adjustment (which is cool), I think SIFT
(XEP-273) does roughly the same thing.  Should we add ping optimization?


On 2/10/10 11:22 AM, "Jonas Lindberg" <jonasl at google.com> wrote:

> Hi,
> 
> Since there were a bunch of interest in mobile optimizations at the xmpp
> summit, I'd like to share a simple presence queuing optimization available
> in the Google Talk service. Hopefully those interested can play with this to
> gain experience in how presence queuing affects battery life and use that
> knowledge when designing future optimizations. This is not a standards
> proposal of any sort.
> 
> The presence queue feature saves battery in two ways: (1) by queuing
> presence going to the client and (2) by increasing the interval for
> white-space pings sent from server to client.
> 
> Protocol example:
> 
> Client enables queuing:
> 
>    <iq from='node at domain/resource' type='set'>
>      <query xmlns='google:queue'>
>        <enable/>
>      </query>
>    </iq>
> 
> Server responds:
> 
>    <iq to=''node at domain/resource' type='result'/>
> 
> The possible 'commands' are:
> <enable/>  -  enables presence queuing.
> <disable/> -  disables presence queuing.
> <flush/>   -  send all presence in queue to client.
> 
> Details:
> 
>  - Queued stanzas are presence available/unavailable. Presence subscribe,
> message, iq etc are not queued.
> 
>  - Queue only keeps the last presence stanza for a given from-jid (i.e. full
> jid).
> 
>  - There is a size limit for queue, when reached presence in queue is
> flushed to the client.
> 
>  - When enabled, server-to-client white-space pinging is reduced to once
> every ~30min.
> 
> Hopefully, some of you will find this helpful.
> 
> Jonas Lindberg
> _______________________________________________
> Summit mailing list
> Event: http://xmpp.org/summit/summit7.shtml
> Info: http://mail.jabber.org/mailman/listinfo/summit
> Unsubscribe: Summit-unsubscribe at xmpp.org
> _______________________________________________



More information about the Summit mailing list