Databases were once the forgotten stepchild of the open-source family. Companies like Red Hat included database software with their Linux distribution disks, but the main focus was on the operating system, the kernel and the graphical interface. A database was just another add-on, like a Minesweeper clone. But now, companies and users are scrambling to realise the value locked up in quality, open-source database software.
The two major open-source databases are MySQL, a lightning-quick database with roots in Sweden, and PostgreSQL, the latest version of a venerable database project (Postgres) that began at the University of California, Berkeley. Both are common programs that are installed on many servers and are often included free with other packages and distributions. Other versions and cousins exist for different markets. Sleepycat Software, for instance, builds and maintains a small database, which it calls Berkeley DB, for embedded systems.
All of the packages come with open-source licenses, an option that often makes them free to obtain, but new options for purchasing support are evolving as companies try to undercut the major proprietary competitors. Everyone is acutely aware of the high cost of packages from Oracle and IBM, and several major players are trying to sell sophisticated support with open-source databases.
Finding high-profile open-source database users isn't hard. Many Web sites, including Yahoo and Slashdot, depend upon open-source databases like MySQL to store articles and comments. Many companies that provide shared servers now offer MySQL databases as an inexpensive option, practically guaranteeing that many low-budget Web sites depend upon the tools.
The products are gaining notice in corporate environments. Robert Beer, a London-based programmer who identifies himself as a "long-term Oracle DBA and recent Postgres convert," says Postgres offers him most of the essential features found in Oracle, with "many of them implemented in a more logical and more ANSI-compliant way."
Still, he grouses that the documentation is scarce and there are few outside options. "I have only found one good book on using PostgreSQL, either in London's excellent bookshops or on the Net. On the other hand, the world is awash with great books and courses on Oracle and SQLServer," he says.
Don't Bet the Farm
Many corporate IT managers share Beer's enthusiasm for the tools, but they often have reservations about trusting their companies' most valuable data to open-source databases. Phillip Moore, an executive director at New York-based Morgan Stanley Dean Witter & Co., says his firm frequently uses open-source products for projects. Databases, however, contain the information that's the lifeblood of his industry. "In the database space today, commercial databases have more to offer for a commercial venture like ours," he explains. "But that's an important caveat. If I was at home, I would be using PostgreSQL or MySQL."
Moore says that in his eyes, many of the open-source tools like MySQL don't offer the same sophisticated mechanisms for juggling requests from multiple users, while products like Oracle or DB2 thrive when faced with these challenges. His company, however, continues to evaluate the products and to consider the best product for every challenge.
In some cases, companies are trusting open-source applications with basic tasks. Stefan Sharkansky, a developer at Personal Fund, says, "I still use MySQL for jobs like storing log files that you don't change. More complex queries, like how much money people have, are handled with Oracle."
Other devotees also praise the open-source software but say the lack of commercial infrastructure makes it difficult to use in a business environment. Filip Hanik, a database administrator, uses MySQL or PostgreSQL for personal and low-budget projects but recommends Oracle for serious projects with big budgets.
"If there's a problem, we say, 'Give me an Oracle consultant.' Then we give them US$250 an hour to fix our problems. We just can't do that with a PostgreSQL person," he says.
Growing Support
Several companies are fighting to change that perception by offering support at more reasonable prices. Great Bridge. for instance, is aiming at becoming the premier supplier of Postgres databases. "The core team [of Postgres developers] has six members. Three of the six members work full time for Great Bridge. Fourteen of the leading contributors are on the Great Bridge advisory board," brags David Mele, the firm's vice president of marketing.
"Support is all we do," he says. "We staff it with experts. It's not a necessary evil for us. We don't tell you to get in line for a feature. You can develop it yourself. We also have a consulting practice. We can help you."
Red Hat is also rolling out its own version of PostgreSQL, which it calls Red Hat DB. Mike Evans, product manager for the group, says Red Hat will target the IT manager who wants "one neck to choke." Red Hat will try to build its business by offering all of the consistency, hand-holding and ease of use of a major brand, at a substantially reduced price. Evans promises that Red Hat will continue to release identical versions on its Web site for download.
"There will always be 30 per cent to 40 per cent that have the desire to get it themselves, and we're not going to stop them. We'll continue to include PostgreSQL in the Red Hat CDs." says Evans. "It really comes down to a shift to what our more business-oriented customers want: 'I don't have the time to cobble it together. You do it for me, and I'll pay you for it.' "Other companies are expanding their open-source offerings. For instance, NuSphere offers a bundle that combines the MySQL database, the PHP scripting language and the Apache Web server. One quick installation ensures that the three work reliably with one another.
"[Open-source software] is such a fast-moving train. We provide at least quarterly updates as part of our advantage product. We keep everyone updated. Some customers say, 'I can build it once myself, but I don't have time to keep up with all the changes.' Knowing that, they always have the very latest [updates] that work together in a supported configuration," says Brit Johnston, NuSphere's chief technical officer.
Johnston's sales pitch aims at the assumption that low-cost software requires more expertise and more time. But many users say such assumptions aren't really fair today. Tools like MySQL and PostgreSQL are very polished and are becoming more so very quickly.
Calvin Dodge, a database developer at Linko Data Systems, says the advantages of slick, polished and professional databases are overstated for everything except small projects. Dodge argues and many other developers agree that the difference disappears once you need to start writing code. All databases, both proprietary and open source, require some time to learn their quirks, he says.
"Access has all sorts of wizards that help one make simple forms and tables but they don't work as well as the complexity increases," he says. "When you have lots of records, you need to go to code-based data retrieval, and in that case, there's no difference in the ease of development."
