by David Strom
(ran in Infoworld 2/10/97)
If you are looking to analyze your web site for broken links, there are many products to choose from, including FrontPage and WebMapper (both now owned by Microsoft and reviewed TK and TK, respectively), InContext's WebAnalyzer and Site Technologies Inc.'s SiteSweeper. This latter product is the easiest of the three products to use, and it can quickly provide a series of reports showing you where your HTML went wrong. That's the good news. The bad news is that it produces other reports that aren't as useful, and it only tells you part of the picture for maintaining your web site.
For example, SiteSweeper has a tougher time showing which pages have different file names but the same content, and it lacks a visual representation of your site's working links and relationships.
SiteSweeper produces its reports in HTML format, like many server analysis tools. There is a table of contents that points towards several reports that have somewhat overlapping information about the links on each page. Probably the most useful report is the one showing pages with bad links. I found several dozen pages with broken links on my site.
Two reports that are worth noting are an image catalog (see screen shot) that shows you every image file it could find and its size. Unfortunately, what it doesn't show you is the path or URL to that image, which would be helpful. Another helpful report is the download size and time for each page, which estimates the time it will take to view the page under various connection speeds.
SiteSweeper does have some benefits beyond link checking, though: it can automatically schedule its reports to run at a particular time of day, something that not all of its competitors can do. But running the reports isn't that difficult, and in my case it took just a few minutes to crawl through my site and gather up the information it needed. This is all the more impressive given that my network is connected via an ISDN line to the Internet.
On the whole, I'd feel better about these products that analyze my web site if I could just get two of them to agree on some simple metrics, such as the number of HTML files and images I have on my web site. For example, neither SiteSweeper or WebMapper could agree on this. The 800 or so pages that WebMapper thought it found turned out to be only 268 individual pages of HTML according to SiteSweeper, and there was another discrepancy over the total number of image files. When I looked at the number of files in NT's Explorer, I got a third set of values. That is somewhat discouraging.
Overall, I liked the organization of WebMapper's reports the best. SiteSweeper would report conditions such as "Internet operation timed out" or "robots not allowed" as errors that I should actually do something to fix. And many of the errors it reported had to do with the fact that my web server www.strom.com is really an alias for the machine soho7.sohonet.com -- you would think the software would understand this and not constantly remind me of this fact.
I tested SiteSweeper on an NT 4.0 Server running on an IBM 365 Pentium PC. The total time from inserting the CD to producing my final report was less than 15 minutes -- that was one of the fastest setups I've seen. However, I wish SiteSweeper would be a bit more careful and not install various DLLs in my
\windows\system directory.
Price: $399
Platforms: Windows 95, Windows NT (4.0 only)
Durham, NC
919 416 3113
919 286 1901 fax
copyright 1997 Infoworld Publishing Co.