Borland announces it will spin off a company to direct the open source code. Psych!

Borland releases code, but changes its mind about the spin-off. Dancing in the street stops.

The spin-off team retreats. A tree grows on SourceForge. Firebird rises from the ashes of angst.

Bring us more candles. That is just the Prologue.
 

Recently, the CEO of MySQL, Mårten Mickos, tipped us off about another important Open Source database project: Firebird. Soon after, we began collecting our early notes on Firebird and its commercial supporter, IBPhoenix, Not long into the project, we could see this was going to be no easy task. In literary terms, anyone expecting Danielle Steele had better prep for Leo Tolstoy.

First, there are those reports about bad odors coming out of air ducts. A fish head FedExed to an employee. A group of startup leaders waiting for the startup that never happens because the parent company drops plans to spin them off. Then come the resignations, forked code, and an InterBase variant under a new Firebird project. Now add to that a guy named Dimitri, billed as the RDBMS guru of Eastern Europe, in Russia promoting a commercial Firebird variant called Yaffil. Next critics question whether Yaffil flies in the face of the Firebird community—but wait, don’t go away.

The story of Firebird is a story of a sound technology and a dedicated community, who are in for the long haul in seeking to place Firebird on the business map. A deadpan explanation of Firebird might go something like this: Firebird is an Open Source incarnation of InterBase, the relational database system, from Borland.

Interbase (company with a small 'b') started in the mid-1980s as Groton Database Systems in New England. Its pioneers were Jim Starkey, who did the software architecture and wrote most of the code, Don DePalma, who wrote documentation, and Starkey’s wife, Ann Harrison. Then the company became Interbase. Ashton-Tate bought Interbase and then Borland bought Ashton-Tate.

Still with us? In 2000, Borland decided that its InterBase 6.0 product will be Open Source under the InterBase Public License, a variant of Mozilla Public License. Associated with this release was a plan to create a separate company called InterBase Software Corporation (ISC) that would direct the Open Source InterBase project. Ann Harrison, the wife of Jim Starkey, was expected to head the new venture and was joined by Paul Beach and others. The code was released, but the associated plan never occurred.

When Borland announced the ISC deal was off, and with InterBase so well regarded as an enterprise-class database running systems at large organizations, news of the spin-off that never occurred traveled fast. Borland itself issued a press release, saying “Borland today announced that it has terminated negotiations with Ann Harrison regarding the sale of the InterBase product line to a start-up venture led by Ms. Harrison." After careful consideration, continued Borland, “We determined that it was not in the best interest of Inprise/Borland's stockholders for us to sell the InterBase product line to a start-up entity which initially would be dependent on Inprise/Borland for funding."

So a rebel corps of developers and users banded together and forked the InterBase code. In July 2000, the aborted spin-off was reborn as Firebird. The Firebird tree went up on SourceForge. To avoid trademark infringement, the InterBase Software Corporation spin-off that Borland aborted became IBPhoenix, a commercial enterprise led by Harrison and others.

“Although we made many attempts to reunite our efforts with Borland’s to strengthen the position of InterBase and its further development, it is obvious that our definitive divorce is inevitable,” reads the Firebird project’s Novice’s Guide. As marked by angst as that site sounds, Harrison tells Open that she now prefers to focus on good things to come. “Firebird is a very healthy open source project with a good future.” She, along with a small team, runs the company, which offers support, services, and consulting for Firebird and InterBase. Meanwhile, Starkey now has his own business, Netfrastructure, which sells a system for developing and deploying web applications. He advises Firebird through one of its mailing lists.

In an interview with Open, Harrison clarifies the past, underscores the present, and comments on drama over Yaffil.

What is the history of Interbase Software Corporation and IBPhoenix?

HARRISON: We got a large investment from Ashton-Tate very early on. The terms of the investment agreement gave Ashton-Tate a minority interest in the company with the option of going first to a majority then full control. Ashton-Tate did exercise both those options, eventually creating the second Interbase Software Corporation, a wholly owned subsidiary of Ashton-Tate.

Shortly thereafter, Borland bought them and chose to integrate Interbase rather than let it continue as a wholly-owned subsidiary. Then they did spin it out as a subsidiary, and brought it back in again less than a year later. When Borland decided to spin it off as an open source company, I created yet another Interbase Software Corporation to direct the open source effort. Then that deal fell through and the last ISC became IBPhoenix, to avoid trademark infringement.

In an InterBase World interview, Jim Starkey says he doesn’t think he could have developed Interbase as an open source project. “I’m an engineer, not a politician,” he told his interviewer. “The Achilles heel of open source projects is decision making.” He says he could not have built a consensus around such a radical new database architecture. Harsh words?

HARRISON: We bicker about this stuff all the time. Decision making is a problem on any project with more than one member. Generally, someone designs and implements a feature and, unless there is a great outcry, it is integrated into the project. Some of the major decisions—porting to C++, releasing an initial C++ version without changing anything that affects the on disk structure, freezing the C code—are made by consensus of senior project members. The discussions are long but so far the decisions have been good.

Starkey does not show great enthusiasm in the ability of Firebird to outdistance InterBase as a world leader in database technology. “I keep waiting for someone to run with it,” he tells his interviewer in InterBase World. “Maybe somebody will.”

HARRISON: Firebird is nearly 20 years old, and it’s a quite conventional database engine. Its design carries some very out of date assumptions about the cost of computing power, memory, and disks. Are we ready to reinvent the whole storage strategy? No. We can, however, innovate in subsystems, as we are now doing with memory allocation routines— comparing implementations and choosing the fastest one that works.

You’ve told me that Firebird’s goal is to cut into the market for SQL Server and Oracle. What makes that plausible?

HARRISON: Beyond price, there’s ease of administration, excellent concurrency, ACID transactions, a rich procedure and trigger language, good support for the ISO language standard, plus of course performance and reliability.

What problems still need to be addressed?

HARRISON: At the moment, we've got a problem on SMP systems with one of our two architectures. The current suggestion is to use the Classic process-based architecture on SMP until we get the thread-based SuperServer running better there.

HARRISON: Yes, some of the people on the project are upset by Yaffil. I'm not one of the people who are upset.

You've said that under the IPL, Yaffil people have a perfect right to charge for their product, as long as they publish their interfaces. So are they publishing their interfaces?

All the Yaffil material is in Russian, so I really don't know. The party being damaged, if there is any damage, is Borland and that company doesn't seem overly bothered. The economic realities in Russia are very different from what we're used to in North America and Western Europe.

The Yaffil group believe that they can build a business on software priced below the shipping cost of software in the west, which makes the phrase "exploited for commercial gain" seem overly strong. The Yaffil developers think that publishing their ideas will allow someone else to sell an even cheaper version of their software and cut into their market. As I said above, economic reality in Russia is different.

When the InterBase Public License was chosen, we balanced the risk of someone deciding to resell the database software against the opportunity to provide database services to commercial applications. We've got a lot of organizations using the license as we wanted and only one that can be seen as taking advantage of it. Seems like a good trade to me.

Can you please clear up what is just briefly mentioned in your interview with InterBase World about bad smells coming out of air vents and fish heads sent by Fed-Ex?

HARRISON: All that happened while ISC was independent. The "sick building" smell from the ventilation ducts was something that seemed to be linked to job grade. People on salary didn't smell anything. We called in an environmental consultant and vacuumed the ducts.

The fish head incident involved two people from support, a very tight-knit group, but also wound pretty tight. One woman who worked late often brought dinner with her. The dinner was often fish and the smell was sometimes quite noticeable. As a joke, more or less, another member of the group put a mackerel inside her monitor, where it became increasingly pungent. She sent the fish head to his house.

So now you feel that the drama is behind you. And there is IBPhoenix to run. It’s more accurate to define IBPhoenix as a profit-making company rather than community service organization?

HARRISON: I only wish it were a profit-making company. You’re right, though, it is a commercial venture. IBPhoenix offers a range of services including premium services and custom training.

Do you expect the number of people in your premium support and service offerings to climb?

HARRISON: Couldn’t change downward. And I think that’s part of the open source culture. Most of our sales are in the lower-priced services like e-mail support and incident packs. People who use open source software think for themselves and they know how to solve problems. They may want to buy a pack of service calls to avoid wasting a lot of time on a minor issue, but they're not going to spend a lot of money on full-time hand-holding. That may have been the problem that doomed Great Bridge. Great Bridge was looking for a quick return.

Will that change?

HARRISON: I really can't predict the future. If open source software or at least open source databases become more mainstream, then, yes, we will start selling premium services. I don't see that happening this year, but it may over five or seven years. Or it may not. IBPhoenix is run by a few very stubborn people who enjoy what they do. We’re here for the long haul.