Welcome to WebmasterWorld Guest from 188.8.131.52
1- This one we know: Accumulated data waiting to be eased in after code change (maintenance).
2- The discovery of new reasons to scrutinize the whole network data for click validity/click fraud reasons, could be a found flaw in current algo getting patched (go to 1), or the emergence of new patterns of suspicious traffic activity, which gets eventually chopped (when in doubt err on the side of advertisers)
This could also explain why some here report losing some of the earnings after such lags (chopped earnings).
I also observe daily mini delay events probably for routine double auditing purposes, mine is around 8 hours into the Google day.
Bear in mind how very tightly connected AdWords and AdSense are - heck, whenever they take one down for maintenance, the other one has to go too (BTW - AdWords maintenance this Saturday, so AdSense will no doubt be down too) So when reports lag, or there's a problem logging in, or it temporarily goes offline for a few, I bet sometimes it's totally something happening on the AdWords side, and not necessarily anything to do with AdSense. And vice versa.
And for all we know, possibly Google Analytics and other products could be affecting it as well. Heck, the whole Google account system, for that matter.
Yes, I imagine much of the work happens using the same MapReduce-et-al framework and CPU pools, and G won't want to have too many spare most of the time as its a waste of capital, space and energy.
1- Code maintenance
2- Click fraud auditing
3- AdWords maintenance
4- Resource redirection e.g. PiP calculations
I would also like to add a couple:
5- Updating the per account SmartPricing variable
6- Re-adjusting traffic conversion probability
(e.g. seasons, event spikes, country specific, topic specific..)
bet on which WebmasterWorld member will be the first to start an "updation delay" thread :)
ROFL, and in all honesty I really have wondered whether they actually do this?
I know how we could stop them...have a permanent "Updation Delay" sticky at the top of the page!
9. Replacement of equipment and failure to back-up/transfer data.
With so many servers they must be replacing lots of them on a daily basis?
[edited by: HuskyPup at 1:34 pm (utc) on June 6, 2008]