EE Times home

News
This Week's News

Profession
The Profession
Career Corner
Ask the Headhunter
Guide to Internet EE
Salary Survey

Columns
Contents
Wintel Watch
Letter from Europe
Institutional Memory
Nano Research Today
Dataport Online
Comm. Unplugged
Weaving Your Web
The Motley Fool

Features
Smart Technologies
Special Reports

Product Info
Product File

Fun Games
Immortal Works
Game Power
Poems

Search

Looking around your company for the next step beyond "brochureware," you are likely to find the most valuable enhancement to your Web site involves posting information about inventory listings, nationwide sales-rep directories, manuals of product specifications, a decade of performance-test results, terabytes of drawings, file rooms full of engineering-change orders, etc.--all collectively referred to as legacy databases (DB). Such databases have been operating on your desktop for years, and often represent the capability statement of your company.

Every database vendor, including the Big Three (Informix, Microsoft, Oracle), have breakneck development efforts underway to produce database-Web site connectivity tools. Many engineering companies understand this opportunity and are struggling to choose one of the competing approaches, to find the $100,000 to get started, and to field something using the beta-quality software that is available.

There is a better way.

Consider that the basic idea behind the leading-edge DB-Web connectivity developments is to arrange an online form to generate an SQL query, pass that to the database, return the report, and convert that into a Web page to display to the user. There are many issues involved with this, including a firewall to provide security to the queried database, but few as important as the ability to scale up to the throughput requirements of the Internet.

Even unpopular Web sites count users in the thousands, and it gets worse. If a call from your Web site to your database goes unanswered, browsers timeout after 120 seconds with a non-specific error message; most users interpret this as meaning that your site is down or defective. The details involve the Internet's basic design, which is stateless (i.e., it doesn't do multithreading or even track user processes). During peak page requests, servers degrade gracefully by parceling out pieces of a Web page to queued users to keep their browsers from timing out. You've all seen Web pages coming to you a piece at a time. Dynamic database queries can't parcel out partial answers.

Business managers know what it's like to be burned by R so you may wish to consider a low-tech, low-risk alternative while the Big 3 conclude their R Consider how much of your legacy data can be navigated using lists of keywords or categories, as opposed to complex SQL queries. If you can provide hierarchical lists of cross-indexed product categories, for example, like Yahoo does for Web-site listings, you don't need a leading-edge DB-Web connection. Your existing desktop database can generate those lists with links to the relevant product pages right now. Those are web pages that are just ASCII text, and your database can do that right now.

It is not appropriate for data that must be real-time, such as stock quotes. If your database updated Web-site changes in batch every night, would that be real-time enough for your application?

If the only way customers can search your data is via complex queries, the offline approach would drown in a combinatorial explosion of lists that must be prepared ahead of time. Keep in mind that most of your customers seek a simple answer to a simple question, and the off-line approach is effective for that. Your challenge will be to forego the brute-force approach of an SQL query, which can cover all problems, and instead analyze your customer's search problem well enough to design the simple, off-line approach.

You will also eat up storage space for Web pages that could be produced on the fly. So? A product page, for example, is typically less than 5-kbytes (drawings and pictures are extra, but you have to store those separately either way), and Web-server hard drives sell for less than 10¢ per megabyte.

If you're looking for a marketing windfall, remember that you can submit each of the pages put on the server from your database to the Web site directories--from Alta-Vista to Yahoo--thus casting a wider marketing net. You cannot do that if the web pages are produced only temporarily in response to a user query.

If you would like examples of operational sites using this simple approach to DB-Web connectivity, take a look at the following:

The online membership management system for the Orange County Business Council, which went from concept to completion in three weeks for under $5,000.

Those fascinated by the technology of dynamic DB-Web connectivity might grouse about how mundane this approach is compared with a fully dynamic DB-Web connection. The proposed low-tech approach does not exclude dynamic DB-Web connectivity; in fact, it compliments it by giving site visitors the option to go either way--as done by all the Web-site directory services. More importantly, your dynamic DB-Web connectivity project can be correctly treated as R while your parachute--the proposed low-tech approach--is up, running, and demonstrating your ability to quickly apply this emerging technology.

Peter Shikli graduated UCLA's School of Engineering in 1972, obtained an MBA from Loyola a few years later, and eventually founded BusinessWare, a company specializing in information delivery, Web sites and Internet marketing. He has hands-on experience designing over a hundred sites, and is the Chairman of the Orange County Business Council's Internet Consortium, You can e-mail him, or call him at (949) 369-1638, ext. 77.

To view past 'Weaving Your Web' columns

To view other columns

---