data sent/received quota in dashboard

[SM]

The dashboard quota chart is telling me I have used over 20MB in both
the sent and received directions in the last 24 hour period. How is
this number calculated? No one besides me is using my app, and I have
used it only a little bit as I am testing. There are only a handful of
files (1 html, 3 png, 1 gif, 1 js) that total about 30k. My app
occasionally does and XMLHttp request to my python script, but the
data sent is never more than 500 bytes and usually more like 25 to 50
bytes.

I'm having a hard time understanding how a lightweight app that is
barely used could have sent and received over 20MB of data in the last
24 hours. I would be surprised if it used much more than 1MB.

FWIW, the app is at www.treksee.com.

The charts:
http://www.treksee.com-a.googlepages.com/gaecharts.png/gaecharts-full;init:.png

Thanks.

[SM]

A quick follow up...

Now when I reload the dashboard it says 0MB for sent and received
rather than 20MB, but somehow I have now used 2% of available CPU. I
find that to be as hard to believe as the 20MB a little while ago. My
app basically just loads a few small files and does a minimal amount
of AJAX.

http://www.treksee.com-a.googlepages.com/gaecharts2.png/gaecharts2-full;init:.png

On a related note, I also occasionally see warnings in the error log
which say an operation that basically just puts a small string in the
datastore used too many resources.

Would it be correct for me to assume these are just bugs in the beta
stage that are being ironed out? If not, I can't see how an appengine
app could be even marginally useful without seriously tripping the
quota limits.

> The charts:http://www.treksee.com-a.googlepages.com/gaecharts.png/gaecharts-full...
>
> Thanks.

[Francois]

I think everyone sees these temporary strange readings in the
dashboard page
they always disappear after a little while
I'm pretty sure they only happen with seldom used apps, the quota
system being optimised for big traffic and not coping well with 3
requests/hour (think divide by zero kind of errors)
anyways, I've never seen it affect my app after it started getting
regular traffic

as for the "too many resources" errors, it's usually caused by module
load being an expensive process on App Engine
the first time your app will strt on a new server, it will load all
modules and take x times more CPU than usual and trip the error
of course, after the first call, all these modules are cached and the
error will not come again
so this also is a problem that does not happen anymore with steady
traffic

[SM]

Francois,

Thank you. That's very helpful information. It never occurred to me
that the system would handle large apps much better than small ones,
but that does make sense.