The flip side is that broken links are a surefire way to frustrate visitors. Rushing to put up pages that disappoint and anger your customers isn't exactly a recipe for success.

Enter the market for automated web site testing tools. Thanks to spider technology, developers can test a site's internal and external links without having to click through them one by one. The spider can also catalog other quality metrics while crawling the site, flagging files over a certain size, pages without titles, images missing ALT tags, etc.

In its flagship product Linkbot Pro 4.0TM, Canadian vendor Tetranet Software, Inc. integrates a high-performance link checking engine with a database, report generator and efficient user interface. The result is a dedicated site checking tool that lets webmasters proof even very large sites (running to millions of pages) without manual labor. Would that fixing the problems Linkbot uncovers were as easy!

Tetranet's challenge is to distinguish its $295 product from the commodity testing capabilities built into page authoring tools such as Allaire HomeSite. Linkbot Pro 4.0 meets this challenge by scoring high on speed, scalability, usability, and reporting options - features essential for maintaining larger sites. In a field considerably more crowded than it was when we reviewed an earlier version, Linkbot remains the best site checking tool Intranet Design has seen.

All the rage

Linkbot is the core of Tetranet's RageWareTM product line: tools that eliminating the "rage" experienced by visitors who encounter broken links, slow pages and unnavigable sites. (Tetranet addresses the navigation problem with a separate product, Wisebot Pro 2.0.) I can report that from the webmaster's perspective, using and installing the software is a refreshingly rage-free experience.

Linkbot can be installed either as an application or as an NT service. The latter can be scheduled to run unattended as a background process, with benefits to both security and productivity. I can see periodic background runs being particularly useful in multi-contributor intranets, where content is fluid and webmasters take a hands-off approach to avoid bottlenecks. By reviewing periodic Linkbot reports, administrators can monitor and maintain web health without intruding on business processes.

Initiating a Linkbot run is as simple as browsing a web site: type in the root URL and hit return. Tetranet has been refining its multi-threaded link checking code since 1996, and it shows in the program's performance and stability. In a test of 1500 pages Linkbot opened 10 concurrent sessions and crawled an average of 150 pages per minute. Moreover, unlike some programs of this type I've used, Linkbot is robust and respectful of machine resources. I had no trouble doing other work while Linkbot did its thing. On the other hand, the multi-threading isn't perfect; Linkbot routinely ignored me when I tried to cancel runs. I was comfortable with this quirk, though, since other spiders I've used tended to crash or hang without provocation.

A crucial feature of enterprise web site checking software is configurability, and Linkbot gets it right. You can exclude parts of a site by depth or by URL, limit scans to internal links only, turn anchor checking on or off and generally tailor runs to taste. Thresholds can be set for flagging "old" pages (those unmodified since a certain date) or "slow" pages (those over a certain file size).

Beyond this, Linkbot Pro 4.0 introduces a number of new quality checks, some more successful than others. One winner is the addition of HTML Validator, the powerful syntax checker embedded in Allaire's authoring tool HomeSite. Another is the ability to inspect the web server's error logs. This information can reveal subtle or impending defects in pages that might not show up elsewhere.

Of less obvious value is a new JavaScript checker that extracts URLs embedded in scripts. The feature only works when the JavaScript programmer uses a consistent programming approach and does not attempt to assemble URLs from their components at run time - in other words, in a scant minority of situations. Also weak is the integration of these new features into Linkbot's otherwise strong reporting capabilities. Speaking of which ....

Read it and weep

While it's running Linkbot displays a screen like the one shown below. This is a useful and customizable interface that can be filtered by object type (web pages, images, PDF files, or executable links), by resource type (http, ftp, mailto, news), or by a number of predefined queries (pages with broken links, old pages, slow pages, pages missing titles, etc.)

Linkbot scan results, directory view. Zoom (30k).

This is not the only output Linkbot provides, however. The tool also produces a series of linked web pages detailing site characteristics. As shown below, this report features graphics showing the relative importance of site defects by type.

Linkbot report, home page. Zoom (28k).

Besides giving managers a rapid indication of site health, Linkbot's reports now track and graph accumulated statistics over multiple runs. This opens the door to historical QA reporting, a cornerstone of continuous improvement regimens. Tracking also provides a means of comparing the performance of multiple webmasters in large sites.

While Linkbot is very effective at discovering site defects, it could be stronger as a repair tool. Webmasters should be able to edit and upload small changes from within Linkbot without using additional software. Neither the program's pathetic text editor nor its integration with Microsoft FrontPage addresses this problem. Adding an FTP client, which basic authoring tools like Macromedia Dreamweaver have done at version 2.0, would make sense.

The wrap

Linkbot's greatest strength is its reliability. It performs well, even on huge sites such as IBM's intranet, which according to Tetranet has over a million pages. Other link checking programs either take prohibitively long or fail outright trying to crawl webs of this scale.

In addition, Linkbot's ability to schedule unattended runs and generate fine-grained summary reports makes it an ideal site monitoring tool. In fact Tetranet Software is piloting a 24/7 monitoring service with immediate error notification via e-mail, pager or SNMP trap. This is a good indication of Tetranet's commitment to the enterprise market, where site reliability equates to customer retention.

Linkbot Pro 4.0 is available for a free 15-day trial or electronic purchase directly from Tetranet Software.

Tetranet Software, Inc.
135 Michael Cowpland Drive
Suite 400
Kanata, Ontario
Canada K2M 2E9
Phone: 613-599-3888
Fax: 613-599-3826