Inflated Page View count using WebTrends 7

[Doug Johnson]

I'm currently working on a project to implement WebTrends across several eCommerce Sites. As we were analyzing our WebTrends licensing needs, we discovered that our load balancing device (BigIP) was creating log entries that are being viewed by WebTrends as Page Views. Following is a sample entry:

"11.9.9.9 - - [12/Jul/2005:00:00:11 -0500] "HEAD / HTTP/1.1" 200
1028 "-" "-" HEAD / -"

As a result, we are paying for an estimated 300 million+ "page views" per year that are being generated by our load balancing device.

The advice we received from WebTrends was to:
A) reconfigure our BigIP to request an image rather than performing a head request.
or
B) Preprocess our log files prior to analyzing them with WebTrends

Neither of these options are very palatable given our current environment and the number of sites impacted.

Has anyone else had experience dealing with this or similar issues?

Thanks,
Doug

[casimir1234]

Although our issues are different..We are having assorted problems with pageviews..One was that we had a redirect that was redirecting back to itself..small error that wasn't even accessible through the site anymore. (By a human anyway basically it was an orphaned file redirecting back to itself.) Sometimes those things happen with a small group maintaining and building huge sites. Well as you know there are many bots out there..this one had a masked User Agent String..It found the redirect and got stuck in a loop. It banged on the page so much that in one day we wound up with an extra 2.5 million page views. -oof.
Sorry about the rambling am in a bit of a rush.

Anyway have a word doc from WebTrends explaining everything about how they calculate page views in regards to licensing. Can email it to you if you like.
-c

[cgrantski]

Ugh BigIP. It creates so many extra lines and a semi-useless cookie too. We preprocess it and all other monitoring hits out of the logs. It can be done readily with cron jobs as part of the overall daily log transport and analysis, so if you're already set up to analyze a complicated array of logs from all over the place, you've done the hard part and this is just one extra small step. (I say this without having a clue about the complexity of your situation so forgive me.) At any rate, we found that we more than made up for the time spent preprocessing by the time saved ftp'ing and of course analyzing.

We have the WebTrends doc on pageview entitlement also; it's worth getting. It clearly says that monitoring hits are going to be counted if they're in the logs.