RE: [lemonade] I-D ACTION:draft-ietf-lemonade-reconnect-client-02.txt

[Randall Gellens <randy at qualcomm.com>]

At 8:48 AM +0000 2/23/07, Ben Last wrote:

>   >From: Randall Gellens [mailto:randy at qualcomm.com]
> > >   Do we expect connections to drop during resync often enough to be
> > >  worth handling it specially?
>
>  Randy: I think that was your quote - apologies if I have misattributed it.

It was actually Philip Guenther, I believe.

>  In our testing of our mobile client, we use a range of mailbox 
> sizes: some are test and some (the larger) are real-world 
> mailboxes.  Like mine, which is embarrassingly large, and those of 
> a group of test users who are evaluating the first release of the 
> client.  We find that connection drop during synchronization is 
> actually fairly common.  Naturally, the chances of losing the 
> connection during sync increase with the size of the mailbox, 
> especially since we find that the client often needs to do a 
> state-comparison sync after first login, and actual cellular data 
> rates can be very slow at times.  We await the availability of 
> CONDSTORE with great anticipation...

Thanks for sharing this data, I think it's helpful, and a point we 
need to keep in mind.

I'd be interested to know which cellular data technology is being 
used, and also what the resynchronization technique is.  And 
naturally, what happens when CONDSTORE and QRESYNCH are used.

>  Also: sometimes even a rare condition is worth special case 
> treatment if the *consequences* of that condition are sufficiently 
> expensive.  A full resync of an inbox with 1000+ email in counts as 
> expensive :)
>
>  Just for interest, some stuff we've found through mobile phone 
> client testing:
>
>  Connection drops occur somewhat more often when the connection is 
> over WiFi.  We're typically finding that mobile phone connectivity 
> over WiFi suffers because of coverage blackspots that are usually 
> not noticed because they're in places where people don't take 
> laptops, because blackspot effects are worse for a phone because 
> transmission power is lower and because the human body is an 
> effective shield when a phone is raised to the head or worn in a 
> pocket!  Since WiFi is effectively free compared with cellular 
> data, though, it's the preferred way to connect.

Very interesting.  Thanks for sharing this experience as well.

>  Aggressive battery-saving (and cost-saving) measures include 
> dropping out of push mode during the night, or when roaming.  When 
> re-establishing a connection after a pushless period it's often 
> necessary to perform a state-comparison sync.

Good point.

--
Randall Gellens
Opinions are personal;    facts are suspect;    I speak for myself only
-------------- Randomly-selected tag: ---------------
Programming Department:  Mistakes made while you wait.

_______________________________________________
lemonade mailing list
lemonade at ietf.org
https://www1.ietf.org/mailman/listinfo/lemonade
Supplemental Web Site:
http://www.standardstrack.com/ietf/lemonade