The Free Database Niche - Response
As President of the Firebird Foundation and CEO of IBPhoenix, I was referred to the article you wrote "Freie Datenbanken in der Nische". Anyway, a friend of mine sent me a translation of the article in english, and we have prepared a response to the article that we would like to see published. I also want to know if it would be possible to publish the original article (in translation) and our response on our Web site.
The Free Database Niche
A response from an open source database supporter.
In a recent article entitled ""Freie Datenbanken in der Nische", Wolfgang Sommergut compared open source database offerings with the products of the big three database companies, IBM, Microsoft, and Oracle. His conclusion was that open source databases were and would continue to be niche products in the near future. We at IBPhoenix see the future differently. Niche markets are early adopters, incubators for new technologies.Open source databases are attracting attention now because they are breaking out of their original niches and threatening the hegemony of the big three.
The goal of open source databases – change the world
The various open source database products inevitably compete with each other in some areas, but they share goals: reducing the cost of data management and thereby improving end-user software. Relational databases are of little direct interest to most computer users. They are tools that other software developers use to make their products faster, more reliable, easier to develop, and more maintainable. The barrier to database use has been the price of database licenses, particularly for low cost software. Open source databases can - will - change the economics of the software market.
Strength in diversity
Each of the big three proprietary database vendors - Oracle, IBM and Microsoft – must solve all problems at all levels so they can lock their customers into their products. Open source databases unlock the power of reliable database management and offer alternatives. An open source database can succeed by solving one class of problem. MySQL is particularly well suited, by design and because of its license, for use in LAMP – Linux Apache MySQL, Php – web applications. MySQL works in other applications, but it dominates in its niche. Others have other strengths. Firebird is particularly suited to be the back end to vertical market products. PostgreSQL is particularly well suited to scientific applications. Sleepycat is extremely low overhead. Ingres has a long commercial history and many high-end features. Cloudscape has the portability of pure Java. Add it all up and open source databases cover nearly all of the database market.
Commercial Light – bait and switch?
Light versions of commercial databases are not plausible as an alternative to open source databases. Does anyone believe that Oracle, IBM, and Microsoft will continue to offer low cost, limited capability versions without the pressure of open source databases? Limited capabilities and limited platforms have limited appeal, even at lower price – especially when the risk of an arbitrary price hike is added to the mix.
Sommergut argues that "Without the backing of large companies, the required database knowledge, necessary for free teams to be able to build up a veritable competition can hardly be organised."
The developers of open source databases are not sixteen-year-old geeks, playing in their parents' attics. Monty Widenius and David Axmark of MySQL built core database technology for 20 years. Tom Lane of PostgreSQL and Michael Olsen of Sleepycat have impeccable academic and industry credentials. One of the Firebird developers, Jim Starkey, designed the Rdb architecture, which Oracle bought and promotes as "a full-featured, relational database management system for mission-critical applications focused on supporting large scale production applications and high performance transaction processing." Open source database projects are anchored by database design heavyweights.
The nature of open source is another strength: the freedom to exchange ideas. We can and do incorporate features and technology from each other. Cooperation is a huge source of expertise unavailable to closed development groups, hobbled by trade secrets.
Large company support
Open source databases are a relatively new phenomenon. They started to attract wide interest only in the past four or five years when MySQL blew holes in the business plans of high-priced web application servers. At about the same time, Postgres moved into the main stream and became PostgreSQL. Then Open InterBase was reborn as Firebird. SAP's partnership with MySQL is first major alliance. Others will follow as companies recognize the benefits of open source databases.
It should come as no surprise to anyone that Microsoft's database is more closely integrated with Microsoft's operating system than other databases, or that IBM systems work well with DB2, or that Oracle tools and Oracle databases are a natural fit. Single vendor solutions give short-term benefits in exchange for the long-term risk of price gouging.
An industry is growing up around the administration, design, and management of open source databases. The products of that industry are cross platform – database and operating system – so learning one tool gives mastery over many.
Sommergut argues that progress is slow in open source databases. The evidence is against him. In the past two years, PostgreSql integrated a versioning engine and added a native Windows port – each a more substantial enhancement than any commercial system has made. MySQL shed its reputation as a lightweight, adding serious transaction support and making huge strides toward SQL-99 compatibility. Firebird's version 1.5 deliberately postponed features in favour of transforming the code base from C to C++. We believe that object discipline will make future features easier to implement, more maintainable, and better suited to the development style of open source projects.
Cost advantage – open source
Sommergut argues that the real cost of open source databases is higher than commercial system, noting that MySQL charges a license fee for commercial distribution and that third-party support contracts cost more than the typical 20-30% of license fees for the lowest level of support available with proprietary databases. Although MySQL and Sleepycat charge license fees for some uses, they are in the minority. Firebird and PostgreSQL can be integrated into any commercial application with no license fee.
The lowest level of support for proprietary systems is maintenance – upgrades to newer versions. Open source upgrades are free, just like the original license. More important, useful, practical support is also free. All open source databases maintain free support lists that provide real answers from people who have solved the problems other users face, and from the developers themselves. How often does Larry Ellison answer a newby question about Oracle, let alone answer it gratis? Tom Lane helps PostgreSQL users nearly every day as do the people who wrote the books, Bruce Momjian of PostgreSQL: Introduction and Concepts and Helen Borrie of The Firebird Book: a Reference for Database Developers. Free support on the open database lists is one of the few things worth a great deal more than you pay.
Companies like IBPhoenix offer fee-based support contracts. We participate in the free lists at the same time we compete with them. Our customers expect, and get, an extraordinary level of support for the rather ordinary prices we charge.
A final price argument is that standard software is seldom certified for open source databases, so your $40,000 per server Oracle license will save you money on Quick Books. That's a lot of copies of Quick Books. Supporting a new database is non-trivial, and third-party developers need time to gain confidence in open source databases. However, offers them more than just low cost. When they add an open source database as an alternate backend, their products run out-of-the box with full capabilities. Moreover, open source databases companies are partners, not potential competitors.
Oracle, DB2, and MS SQL Server are right to feel threatened, and we'll see a lot more FUD (fear, uncertainty and doubt) from them on this subject in the next couple of years.