Tuesday, June 20, 2006

For those testing/using Direct Push, there is a log file created
on each device that records all the information related to the sync
process between the device and your server. The file name is 'ctrllog'
and is located in the \windows\activesync folder. One useful bit of
information from this file is the current heartbeat interval your
device is using. Values are in seconds with the device starting at 480
and increasing the interval as conditions warrant it. On my device,
the heartbeat tends to hover between 780 and 1080.

This value is critical in the consumption of battery power so if your's is hovering
at 480 most of the day your battery will most likely be consumed quickly.
Just an FYI... they announced at Tech Ed that E2k7 Beta 2 would be
available in July to the general public.

Another intersting note mentioned in one of the sessions.... the log
file size has been changed from 5mb to 1mb to help minimize the chance
of lost data when using LCR or CCR. Looks like we will have a bunch
more log files now.

Monday, June 05, 2006

Here's some interesting info I've found relating to direct push. In the KB
article below it states that the default setting for 'maxheartbeatinterval' is
45 minutes. So if nothing is set in the registry, this is supposed to be the
value used. We have seen a large increase in battery consumption when upgrading
our devices to a rom with direct push.... in the nature of 30-40% reduction in
run time. This didn't make sense to me considering how often the device was
syncing (about 140 messages a day, maybe 10 minutes of voice time) so on a hunch
I manually set the 'maxheartbeatinterval' to 45 minutes. Since making this change,
my battery life has gone from 10-12 hours per charge to 24-26 hours per charge.

I've verified this change on two different devices (a PPC-6700 and a 700W) so
there must be something to this setting. It may be that the setting is forcing
Exchange to use this setting irregardless of the network conditions but I have
seen no adverse effects of changing this. If you have seen similar decreases in
runtime, try this change and let me know your results.


Saturday, June 03, 2006

Here's some good documentation on V2 of the IMF. Much improved over
V1, we are actually using this as an additional layer of antispam filtering
and have seen good results.