To answer your questions:
On 05/24/2010 04:28 PM, Daniel Pastore wrote:
After these 3 tasks, when will the statistics be recorded?
Few minutes after the upload finishes?
Our web servers will see the download.eclipse.org/stats hits
within minutes, but for performance reasons, we import the current
day's hits at midnight (Eastern time). In other words, today's
downloads will be available tomorrow for querying.
Another question I have is if the stats server detects
that the same request is coming in a short time from the same IP, so it
doesn't count every hit from my company network, for example.
There is no intelligence implemented here, since multiple hits
from the same IP could possibly represent multiple (legitimate)
downloads. I think the key metric to any download statistic is the
trend information -- not the actual number.
Keep in mind -- we will see ballot-stuffers. It's only a matter of
time. Your trend data should make it obvious, however, so it should be
easy to normalize.
Thanks,
Denis
|