Who's online page views decreasing??

Good day,

"..Page views get to about 30 then start to decrease..."

This is not a problem, this is a "by design behavior". In fact this figure doesn't really represent the people actually online, but actually, how many people that issued a hit on your site within the period of time set by the "SessionDuration". I personally changed the way the information was reported to have a more "realistic" value as this approach wasn't "accurate enough" to my taste.

I'll have to look back at the query that was initially supplied by MetaTraffic and verify the changes I did if you want me to post the way I'm doing the counts in that regard.

"...Traffic Summary report Visits is always on average double (sometimes slightly less) the number of Unique Visitors..."

If I recall well, the "visits" are calculated using the IP address and the SessionID. So for the same IP address, if you have 2 session IDs in your DB, this would normally indicate that the user did not disconnect from the internet, or at least his IP leased has not expired, and he came back to your site later, thus having IIS creating a new SessionID as the previous one has expired (as per the server's setting [Session.Timeout]), but still with the same IP address.

The visitors are based uniquely on the IP address, which might not always be accurate, however there are not a lot of ways to uniquely identify a user, so we have to go with what is available.

What will disrupt the logic, is the users accessing the site using an ISP via a proxy that dynamically change the IP address on the fly, such as AOL. Metatraffic can report 8 users online, however this is the same "poor" user, ("poor" simply because they are dealing with AOL, no offense to AOL users ;)) for which the IP address has been changed dynamically 7 times within the last "SessionDuration" minutes.

"..If I set it to say 10, if the user is stil browsing after 10 minutes will they still show up in the Who's online report?"

Yes it will still show as an online user since a hit was performed within the last "SessionDuration" minutes scope specification.

To minimize the calculation errors induced by the "SessionDuration" set to an arbritary value, I would strongly suggest you to set it so that it will match the server's "Session.Timeout" value.

The reason for this is simple, if you set "SessionDuration" to 60, as per the defaults if I recall well, a user is considered online as long as a hit was detected from him within the last 60 minutes.

However, IIS default is 20 minutes and the user might have left the site, however as per the 60 minutes, he still show as online, which obviously might not be the real situation. It must be understood clearly that a session times out after the Session.Timeout has elapsed without any sign of activity from the user, it does not indicate that a session will automatically end after Session.Timeout minutes.

Finally, if even after doing some adjustements the figures are still not showing accurate values, perhaps changing the "Server.Execute" for a basic SSI might minimize some problems.

Some friends of mine noticed differences in the overall situation when switching from one method to another, although they haven't provided me with all the details, so it might be worthwhile to consider giving it a try.

I personnally use SSI, would it be IIS4 or IIS5 for my tests, thus allowing be to evaluate the scripts on both servers. Server.Execute came in the picture with IIS5 and up.

I understand that IIS4 is old, however quite surprisingly, there are a lot of hosting services that are still using IIS4, specially the free ones, which are great to do some tests.

Hope that clarified the situation a bit more for you.

Happy tracking! :D

Daniel

Thank you for your detailed reply.

Mal