Classic Hits, Part 1
Hit counters are the newest Web craze; which ones will rise to the top of the charts?
By Steven J. Vaughan-Nichols
By now, you probably know how to record site hits, and that there are at least half a dozen cheap counter programs available that serve the purpose. What you may not know is that counting raw hits to your site is next to meaningless. What you really need is a program that collects all the information it can get from every visitor to your site. With that information in hand, the software must organize data that's useful to both you (the Web administrator) and your company. One or the other is fine, but you really need both. For example, you might want to know when the site is most active, while your marketing manager wants to know who's spending the most time on a given page.
Simply delivering raw statistics, however, isn't enough to make a good product. Raw data isn't tasty at all. The information must be seasoned by a proper analysis tool before it is useful. A second-rank hit analysis tool will give you its own internal report and graph generator (reports will often be generated as HTML), but you may already be using a host of other database and reporting programs. Who wants to learn how to use another reporting software package when your company has already standardized on, say, Informix for its DBMS and Crystal Reports for reporting?
A first-rate traffic analysis program should be able to deliver reports to you, but it must also deliver data in a form that other programs can use. The top programs will give you a way of setting up automated data massaging to get the data primped for analysis, and then use Open DataBase Connectivity (ODBC) or another real-time data transaction protocol to deliver the data transparently to your company's DBMS and report programs. Short of that, the program should be able to export its data to such popular DBMS formats as dBase, FoxPro's DBF, Lotus' WKS, or non-proprietary formats like DIF.
Beyond that, you also want a program that will work with your existing log files. If you're like most of us, you want workable information now, not in a month when you'll have a representative data set to work with. In practice, you want a program that will work with both the Common Log Format (CLF) and Extended CLF (ECLF). That should take care of most modern Web servers.
Finally, you need a program that won't choke on enormous log files. Even if you allow your log files to grow to jumbo proportions, a good program will at least let you access the entire log. A great one will let you extract only the sections of the logs that you want to examine. You'll also want software that's fully TCP/IP-compatible to avoid operating system clashes.
Are there such paragons of Web-counting virtue out there? Nope. Oh, with a crack team of C and Perl programmers, you could put together a homebrew that would do it all; but few of us can boast such astounding resources. Instead, let's take a look at what is commercially available.
[Move on to the next part of the article]
Web Developer® Site Feedback
Web Developer®
Copyright © 2000 internet.com Corporation. All rights reserved.
Database Journal: The Knowledge Center for Database Professionals
