Aug 25, 2007 (02:08 AM EDT)
Open Source PostgreSQL Developer Bruce Momjian Spills The DB Beans
Read the Original Article at InformationWeek
In a recent interview, the lead integrator and bearded sage of the PostgreSQL project, Bruce Momjian, talked about the emergence of a commercial company, EnterpriseDB, behind PostgreSQL and the vitality of the database itself.
Momjian is a database architect employed by EnterpriseDB and often spends a day out of the week at the Iselin, N.J., company. But the self-taught programmer mainly works on coordinating open source PostgreSQL development.
EnterpriseDB produces the commercial product, EnterpriseDB; its Advanced Server retails for $5,000. It's based on PostgreSQL with commercial add-ons.
Momjian said the launch of EnterpriseDB has helped the visibility of PostgreSQL. As open source code without a company, it's had to spend several years putting its house in order, achieving a high measure of SQL query language compliance before offering itself as a robust, full-featured database system. PostgreSQL emerged in 1996 from Postgres, a project at the University of California at Berkeley. The Berkeley system used the PostQuel query language instead of standard SQL.
"We need someone saying we're a lower cost, disruptive technology," Momjian said. Few personalities, including Momjian, have stood up inside the project and made that assertion, the way Marc Fleury did for JBoss. PostgreSQL is something its developer community tends to leave for the world to discover for itself.
Momjian in some ways is the opposite of Fleury. He's established himself in the midst of a sophisticated database project and is happy with that. He has no aspirations to lead a company. Fleury built a company around his central role at JBoss.
Momjian is glad, however, that a commercial company is active in the project and in promoting the merit of the PostgreSQL work. EnterpriseDB values the community and works through the processes it controls. It also "seems to understand where the value-add is for its own product line -- the Oracle compatibility and the one-throat-to-choke idea of support," he said.
EnterpriseDB also employs a staff of programmers who work on features for PostgreSQL. As full-time developers, they sometimes submit big additions for review by the project's core committers. "It can be hard for someone who works as a volunteer every weekend to review 10,000 lines of code," he notes.
But the large community around PostgreSQL -- its e-mail lists go out to 51,788 recipients -- is good at inspecting and hammering on large additions. It's got the ability "to eyeball the ideas in a big piece of code," he said, test them, and help filter out the good from the bad.
The community had gained practice in filtering big ideas in large additions when EnterpriseDB came along, he said. Additions such as transaction semantics, replication, failover, and recovery, "are really hard features to add to a database."
PostgreSQL also is different from the most successful open source code database system, MySQL, which is more like JBoss in representing a project that resulted in a company forming around it. MySQL was a system that early on sought to rapidly serve read-only data for purposes of building Web pages, then added on the dynamic relational features of updating or deleting data.
PostgreSQL sought to match major relational database systems and afterward addressed performance. "We didn't focus on performance until the end. MySQL had that right at the start," he acknowledges.
But Momjian thinks the developer community around PostgreSQL is its greatest strength and he wants to see it drive the database system as a competitor against the strong systems built by single companies, whether open source or commercial. That will be an uphill battle.
"If the community model isn't as good as the inside-the-company model, then PostgreSQL won't do very well," he said. But he's betting his efforts that PostgreSQL will gain momentum and win a larger place for itself in the market.
Momjian is the author of "PostgreSQL: Introduction and Concepts" (Addison-Wesley, 2000).