I'd been planning a write-up anyway on the very different open source strategies of PostgreSQL-based EnterpriseDB, PostgreSQL-based Greenplum, and PostgreSQL-based Aster Data – not to mention more-or-less-PostgreSQL-based Netezza, Vertica, and ParAccel, Ingres-based DATAllegro, and MySQL-based Infobright. Then Infobright decided it was taking its analytic DBMS engine open source too – a very worthy move in its own right -- making the subject particularly timely. So here goes.
There are three basic ways open source can manifest itself in database management products. First, a DBMS can truly be open source. Products that can be had in true open source community editions, usually GPLed, include MySQL, PostgreSQL, EnterpriseDB's Postgres Plus, Infobright (in connection with MySQL), and a number of others. All the ones I mentioned except PostgreSQL come from specific companies that also offer enterprise/professional editions, on subscription, with or without additional closed-source features. (And EnterpriseDB/Postgres Plus is increasingly stepping into that role for PostgreSQL.)
Second, a wholly closed-source product can just incorporate some of an open source project, as a kick-start. That's what a lot of vendors have done with PostgreSQL. Greenplum has played a bit with giving back to the community – Bizgres and so on – but basically just took PostgreSQL and built a commercial product by much extending it. That's what ParAccel and Aster Data did too, but with even less attention paid to the PostgreSQL community. Meanwhile, Vertica used PostgreSQL as a design standard; so as to be compatible with lots of tools, it built a PostgreSQL-compatible product that didn't actually include any PostgreSQL code. And DATAllegro and Infobright built products in connection with the products and companies Ingres and MySQL respectively, before DATAllegro was bought by Microsoft and Infobright decided to go whole-hog open source itself.
Third, a DBMS – probably closed-source – can rely on non-database open source technologies. At their core, DBMS are big SQL interpreters. But for years, they have interpreted other languages as well, including Java, Perl, Python, and R, all of which were mentioned in the recent MapReduce announcements by Greenplum and Aster. Also common is for an appliance vendor to run a custom flavor of Linux, something that software-only DBMS vendor Exasol does as well.
Bottom line: Open source software is hugely important to database management, most especially in the vibrantly innovative market for high-end analytic DBMS.
Other posts today on open source DBMS
Curt Monash is a leading analyst of and strategic advisor to the software industry. Praised by Lawrence J. Ellison for his "unmatched insight into technology and marketplace trends," Curt was the software/services industry's #1 ranked stock analyst while at PaineWebber, Inc., where he served as a First Vice President until 1987. He subsequently co-founded Evernet, Inc., a $40 million networking systems integrator. Since 1990, he has owned and operated Monash Research, an analysis and advisory firm covering software-intensive sectors of the technology industry. In that period he also has been co-founder, president, or chairman of several other technology startups.
Curt has served as a strategic advisor to many well-known firms, including Oracle, Microsoft, SAP, AOL, CA, and Netezza. Curt earned a Ph.D. in mathematics (Game Theory) from Harvard University. He has held faculty positions in mathematics, economics and public policy at Harvard, Yale, and Suffolk universities.
The Leader in Network Knowledge
Another way open source DBMS's can interact with companies
There's another interesting aspect of the first way you mentioned. A for-profit company with real revenues can hire developers to improve the open source database (contributing their work back to the open source community). Even better, they can hire the people who are current top developers of the open source database.
For a long time, Python was maintained primarily by Python Labs, which was a part of the Zope company, which depended heavily on Python. Six developers (including Jeremy Hylton) were paid by Zope to work half-time on maintaining Python, and they were the core maintainers, with (I think) about 28 open source contributors outside of Python Labs.
Many open source projects/products that get to be really good and popular end up being supported by companies, with the most obvious case being Linux (e.g. IBM).
In the case of PostgreSQL, my understanding is that EnterpriseDB has taken this approach, although I do not know the precise details.
Greenplum Closes but No Cigar
Bottom line: The issue with these companies is that they do little but pay lip service to the open source community. Instead of giving useful innovations, they keep them closed and kill off real interest/momentum. Prime example: Greenplum bizgres is essential dead. The site's (http://www.bizgres.org/home.php ) most recent news is a User Group Meeting on June 9, 2006. I guess they didn't have much turn out and/or didn't care to have a single piece of news on bizgres for the past 2 years.
Companies like ParAccel, Aster Data, & Greenplum have stolen from open source DBMS and not given back sufficiently. Now it is too late as far as a strong business model. The days of thriving open source DBMS innovations are coming to a close with SUN buying / smothering / slowly killing MySQL. (The founder of MySQL just quit.) DATAllegro belongs to MSFT. Even closed-source Teradata is rumored to be bought by SAP.
It is frustrating to see this abuse and lack of support/giving to this potential hotspot for dbms innovations.
Teaming and Collaborating On Open Source
PostgreSQL is gaining significant momentum in the market and is primarily driven by strong collaboration between EnterpriseDB's team and the terrific PostgreSQL committers and a few other companies that deliver code and insight/expertise back to the core PostgreSQL engine. A good example is Tom Lane of Red Hat, a bright and talented database guru who does a great deal to help PostgreSQL efforts and people who are trying to better understand capabilities of the engine. Just look at the PostgreSQL discussion lists. You will see Tom frequently giving back and he does an excellent job. Others adopting PostgreSQL in their underlying solution, you know who they are, very rarely if ever give anything back to the community that I"m aware of. That's just the way it is. They take a snapshot, they copy code, whatever. The reason, its based on BSD and therefore companies mentioned in this thread have a great deal of freedom to use PostgreSQL as an underlying database engine. It's their choice if they want to give back or not. Those that do give back and actively participate enjoy working with what I feel is one of the best, long term, open source DBMS projects on the planet. Yes, I'm biased, but I've also been in the data management space for a very long time.