Even that tired old saw about page locking versus record locking won't go away. The good folks at Sybase added support for record locking--except they did it about five years after everyone else did. The net result is not only isn't Sybase keeping up with other vendors' proprietary extensions, but it isn't keeping up with the rest of the industry--a sad fact considering Sybase once set the bar for standards.

The View From Here My point of view is as a humble participant in the IT trenches. We take the products we get from vendors and we manage to make them work. In the end, Sybase basically has good technology, and it's a reliable business partner. Meanwhile, wind and weather permitting, maybe it will glean a clue or two from this column and start whistling a new tune.

Unless Sybase changes its ways, the company will be relegated to living off its early success--its installed base. Furthermore, unless the core database infrastructure can succeed in its own right, Sybase's ability to sell other products into that installed base will suffer. And user confidence will falter.

Why should I believe that its Jaguar is the solution for transaction processing and will function as an ORB (object request broker) when its lack of record locking still causes page locks?

Why should I believe that its PowerDynamo is the answer to dynamic SQL when the optimizer loses its way and "forgets" plans on queries in large tables?

And what can shake my impression that Sybase is stuck in the early 1990s when Oracle is about to introduce an imbedded Java Virtual Machine in the database and I can't even have a user-defined function in Transact SQL?

Whatever Is One to Do? There are three small steps the folks at Sybase can take to turn this around before it's too late.

· Core Competencies First, the company should concentrate on its core competencies. The database server and its immediate supporting infrastructure like Replication Server need to come first. If Sybase doesn't compete on the feature level, then it should compete on depth and lower labor costs; it still takes too much talent to run a shop based on its products.

· Linux Support Second, Sybase should support Linux fully and support it for next to nothing. It will cost the company very little in actual dollars, yet it will gain the support of ISPs and other die-hard Unix users, which will serve it well in the long run. Sure, Sybase's partners in the RISC server world will be annoyed, but what contribution is that making to its overall market share?

· Improve Internet Licensing And finally, Sybase should come up with a better Internet license. The folks there have said if customers want to drive their Web sites with Sybase, they need the equivalent of an unlimited user license--not simultaneously connected users, but unlimited users. To add insult to injury, every time customers ask about price, Sybase salespeople probably give them two different answers. It's an easy plan for getting ahead: The Internet is mature, so Sybase should come up with a correspondingly mature licensing policy. Simply stated, its customers cannot predict what licensing costs to plug into their business model, and Sybase can help them avoid doing that by implementing a realistic, simply stated, aggressively priced license.

The Immediate Danger Sybase is in danger of becoming irrelevant. Irrelevant in the same sense as Borland. Borland, like Sybase, creates decent products, in some cases even better than its competition. However, both companies' grasp of the mind share of the developer and the customer is slipping. Sybase probably can continue like this indefinitely, slowly losing both market share and margin.

To turn itself around, Sybase will need to exhibit two traits that have been missing lately: focus and imagination. Focus to concentrate on SQLServer and related products Imagination to try something new. The recipe to turn this around is simple: Stick to one's knitting by unifying efforts on core functionality, go against conventional wisdom by supporting Linux and prove yourself to be a better partner by implementing a better Internet license.

It never hurt anyone to stick to one's knitting, but Sybase, my friend, you must change your pattern of behavior in order to be a better partner.

Brian Walsh is the founder of bwalsh.com, Portland, Ore., a networking and communications consulting firm specializing in Internet and client/server product strategies, development and testing. He can be reached at www.bwalsh.com.