At first glance, the story of Ingres sounds all too familiar. It’s the beta vs. VHS story of betting on the wrong technology. It’s the story of the company that suffered ham-handed marketing. And it’s the story of the company whose assets got buried as collateral during a string of acquisitions.

At second glance, the story sounds more redolent of the lyrics from an old Grateful Dead tune, “What a long, strange trip it’s been.”

Dismissed for years as legacy technology, Ingres managed to hang on to 10,000 customers across 58 countries. Maybe the size of the installed base is not surprising, because if your company has committed to a database, until death would you ever consider parting. That’s why you rarely see new database customer “wins.”

CA, which owned the company for roughly a decade, invested real R&D until divesting the company last year as part of change in corporate direction. And it “fixed” the problems, such as lack of row-level locking and SQL support. And, while the database market has often been dismissed as a mature sector, there are numerous opportunities beyond the obvious targets of mission-critical transaction systems in Global 2000 companies, which are not about to be displaced. How else to explain the open source database market, which has roughly a half dozen suitors?

This week, Ingres senior vice president of engineering Emma McGrattan spoke about how a forgotten company found new life with open source. The challenge was that after a decade of serious R&D, which brought Ingres release 3 to scalability and functionality in the Oracle 10g league, the product still suffered a perception gap as a CA orphan on life support. And even the idea of releasing a free version on Linux wasn't terribly compelling, given that everybody except Microsoft was doing same. As for Microsoft, they did the free, just not the Linux.

That was the point when the unit decided on its Hail Mary pass: make the product open source.

Of course, anybody familiar with the travails of companies like Sun would understand that there may be lots of legal ramifications to going open source. And that provied one of the biggest hurdles. Initially, CA (which still owned Ingres) tried its own “trusted source’ license, to underwhelming response in the community. Post-CA, Ingres redrssed matters by releasing the new version, Ingres 2006, under the more familiar GPL. Of course, when you offer code that is open source, it better not include code that is owned by somebody else, because you don't want to get caught in another SCO-type lawsuit. CA used Black Duck Software to check Ingres source code for IP.

Then there was the matter of cleaning up the code base. Just as you tend to clean your house when you have guests, software companies preparing for open source must also tidy up. Because, if you open this to the community, you had better clean out the obvious blemishes. In this case, that involved fixing up the documentation and replacing hacks or workarounds with more robust bug fixes.

But here’s the kicker. Like a virtual locker room, code not intended for the light of day must be cleansed of any customer references and, surprise surprise, vulgarities. “Some rude words got into the source,” admitted McGrattan.

Then there was the spin-off. Ingres' initial forays into open source (code cleaning and all) initially happened under CA's watch. CA even sprung a “million dollar challenge” for third parties to develop porting tools. And then after all the hooplah, CA's management changed (facing legal issues, the company had bigger fish to fry) -- and with it came a change in direction to Enterprise ITManagement (EITM). The new team under John Swainsson decided that databases were not part of the EITM focus.

Spun off, the company is fighting a new perception battle with powers like Oracle who insist that the market should be measured in license revenues. Of course, based on that criterion, upstarts like JBoss should be no match for IBM or BEA. But moving to a mainstream open soruce license (GPL), Ingres could at least restart life already having a critical mass customer base.

The story is hardly over. Having been spun off by CA last year, Ingres is now holding its coming out party. It has released Ingres 2006, which it claims is the most scalable open source database out there. It has unveiled a reseller program that gives its partners more leeway than if they were working with an established player like Oracle. Its existing base is paying maintenance under existing contracts for pre-open source versions of Ingres. Ironically, one measure of Ingres’ success may be the degree that its existing base migrates to the open source Ingres 2006, which in some cases might mean lower cash flows.

Probably more importantly, with the emergence of open source, industrial strength databases are now becoming available for the rest of us. The continued growth of the web tier means that there are empty spaces for new database products to fill.

In other words, the market at the low end has opened up again.

But there's obviously plenty of competition, with open source database becoming literally a dime a dozen. There are roughly a half dozen or so open source rivals, not to mention the traditional proprietary folks whose “Express” entry level products are free or almost free and have similar ‘community” level support. In its first iteration, Ingres made the wrong bet on technology. It had marketing that earned the ridicule of Oracle’s Larry Ellison, who called Ingres the company that would push sushi as dead fish.

This time around, Ingres like any open source player requires a dose of social engineering to get the community excited. Who knows? Maybe this time around, dead fish could get trendy.

---