Question about Missed Hits

[Mickey_TLK]

Hey Rich

For months my hit count numbers on all of the "top sites" and the numbers my server is showing have not matched up. Can anyone help me understand whats going on with this, and fix the problem in my layout, ect that might be causeing the hits not to register? For example the "boaphile" hit count shows me as

Average Today Yesterday 2 Days Ago 3 Days Ago
Unique Pageviews 64.25 43 54 76 84
Total Pageviews 363.25 236 266 415 536
Unique Hits In 2.00 1 2 3 2
Total Hits In 2.50 2 2 4 2
Unique Hits Out 1.50 0 3 2 1
Total Hits Out 1.50 0 3 2 1

But my server shows my hit count as

Unique pageviews 134.93 78 118 172 142
Total Pageviews 1319.50 360 1093 2695 1484

Its been like this for months now, on all the counters. As such I have to believe either my servers count is wrong, or something is set up incorrectly on my site so the hits are not registering. Is it possible that my site takes to long to load on slower connections, as such the hits get missed by the top sites counters?

 

[WebSlave]

No really sure, Mickey. But one thing you have to bear in mind is that topsites programs put a pretty significant load on a server. Here's what happens when you have a topsites graphic on your site when a user visits a page. The user's browser downloads the html script to be displayed on their computer screen. When the banner graphic code is interpretted by the browser, this sends a signal to the remote server where the topsites code resides with the code of which site the code is coming from. This triggers the server to load the incoming hit module, which calls up the MySQL database and queries the database for that site ID to see which graphic (depending on their rating in the list) to send back to the calling system (yours) and also updates the database with the hit that your request triggered.

Now this doesn't sound very tough when it is broken down into small steps, now does it? But imagine a topsites program with hundreds of listed sites and those sites getting anywhere from 1 (the zeros really don't count, since they don't have the banner graphics on the remote sites) to 50,000+ hits per day. Each of those hits triggers the above mentioned action within the topsites server.

Now a programmer has a number of choices when programming something like this kind of program in order to accomodate loading problems. The two most common ways are to discard the hits when the server is too busy doing it realtime, or to store them in a buffer and let the server process those hits as batches of data. Real time systems will bog down during heavy traffic periods, and most likely drop substantial numbers of hits. The buffered method tends to work better, but the tradeoff is that the numbers won't be realtime at any given time someone views those statistics. Both methods obviously will have numbers that differ quite a bit from what you will see on your own local website. Your local website counter simply has to count ONLY the hits incoming to your own website, which is much easier to do in real time.

Of course this is all simplified somewhat and I am speculating on some parts of it. But from what I have seen of the way my own topsite counter works on FaunaTopSites and comparing them to the other shareware code that some other sites are using, I think the code on FaunaTopSites is much more accurate. There was in incident a while back when a web site listed on FaunaTopSites got listed on a high profile webpage and they experienced a tremendous number of hits on a daily basis. I talked with the guy running that site (TurtleTimes, I believe) and he said that the FaunaTopSites script was the ONLY topsites script that was able to keep up with the hits and give an accurate accounting of the hits they were experiencing.

Anyway, hope that helps a bit.