[jdev] Re: XMPP Ping method?

Dave Cridland dave at cridland.net
Fri Nov 3 04:39:10 CST 2006


On Fri Nov  3 01:49:06 2006, Remko Troncon wrote:
> Not that I have experience with mobile devices, but if you use 
> zlib,  the overhead of doing a ping should reduce to one byte plus 
> a few  bytes of padding every call in every direction. If you do a 
> ping  every minute, this bandwidth overhead is neglectable compared 
> to the  bandwidth overhead of presence updates alone. If you do 
> polling and  send/receive more than just the ping in one request, 
> the impact of  the extra bytes of padding diminishes as well. A 
> while ago, someone  claimed that zlib on mobile devices was not a 
> good thing, but I'm not  sure if there was proof of that.

All the handset developers we talked to in Lemonade strongly 
suggested that there was virtually no way that any power consumption 
through the CPU used for zlib could possibly outweigh the power 
consumption for transmission.

That said, you absolutely don't want to be doing a ping every minute. 
About every 4 should be more than enough, and ideally, you'd just 
have the server transmit - you don't want the client to be 
transmitting at all if you can help it.

Dave.
-- 
Dave Cridland - mailto:dave at cridland.net - xmpp:dwd at jabber.org
  - acap://acap.dave.cridland.net/byowner/user/dwd/bookmarks/
  - http://dave.cridland.net/
Infotrope Polymer - ACAP, IMAP, ESMTP, and Lemonade



More information about the JDev mailing list