Users take open source databases for a spin

A growing number of network executives are finding that open source databases are reaching a state where they can become the latest addition to their inventory of open source tools.

No one so far has ripped out a clustered Oracle 9i database on multiprocessor Sun servers and replaced it with a free, open source database downloaded from the Web, running on a bunch of Intel-based Linux servers.

But a growing number of network executives are pioneering these freely available data managers. They're finding that open source databases are reaching a state where they can become the latest addition to their inventory of open source tools, including the Linux operating system, the Apache Web server and the Tomcat Java servlet engine.

Users say the attractions include:

  • Very fast performance, especially in read-only applications.
  • No or nominal, licensing costs.
  • Low administrative and operational costs.

Burlington Coat Factory Warehouse is a Linux user looking at the open source database technology. The Burlington, N.J., company deployed Linux servers in its retail stores nationwide a few years ago.

"We're not sure about embracing open source databases yet," says Brad Friedman, the company's vice president of information servers. "But we will probably do so in our retail locations. As to the back-end servers, we're still ingrained with Oracle, which has a fair amount of support for Linux."

Friedman's comment shows a typical pattern in companies that are weighing, or experimenting with open source databases. High-volume database updates, which are the essence of transaction-processing applications, remain anchored on products such as Oracle's 9i and IBM's DB2 Universal Database, and increasingly Microsoft's SQL Server. But there are a host of new application areas that don't require the complex, and expensive, features found in these massive products.

"Some of our groups that use Oracle never were wedded to Oracle's high-priced features," says Jeremy Zawodny, technical Yahoo, for Yahoo, Inc., in Sunnyvale, Calif. "If you don't need these features, you probably don't have to pay the price for them [with an open source alternative]."

At Yahoo, the MySQL open source database, from MySQL AB of Sweden, has "spread from being used by a few groups to the core infrastructure" of the Internet portal, Zawodny says. MySQL handles vast amounts of data from hundreds of daily news feeds, such as Associated Press and Reuters, stock market tickers and so on. The data is stored, marked up and posted in Web pages in response to mouse clicks by Yahoo visitors.

"MySQL is a core piece of that content-generation system," Zawodny says. "This alone accounts for a great deal of the content we publish on a daily basis."

Open source makeup

Open source databases typically are available for free or a nominal charge.and include complete source code. Finally, in accordance with the terms of the GNU General Public License (GPL), the users typically have the freedom to change any part of the source code, and use it without charge, as long as they publish the change. Once published, the change can be used by anyone.

An alternative arrangement is the Berkeley Software Development (BSD) license, which is used by PostgreSQL.org. Developers can "use, copy, modify, and distribute this software…for any purpose, without fee, and without a written agreement…." According to one developer working with this license, it lets him incorporate the database into a software application that he owns. In other words, open source code can become proprietary product under the BSD license.

There is an array of open source databases. Firebird, based on Borland's venerable InterBase database, is one of the few that have the support and blessing of vendors or well-organized communities of coders.

MySQL also is proving popular in the open source community. "The open source world likes MySQL," Yahoo's Zawodny says. "Every time a new [programming] language comes out, usually the first thing developers do is add database connectivity to MySQL. There are bindings and drivers for all kinds of languages. You don't usually see this [response] with a closed commercial product."

PostgreSQL is the most mature of the open databases, and PostgreSQL.org maintains an extensive and well-organized Web presence for its developer community. PostgreSQL, Inc. is a Canadian company that offers applications along with support services. Red Hat bases its product offering on PostgreSQL.

A storehouse of innovation?

The open databases are often storehouses of innovation. MySQL for example has won praise for an architecture that has a core relational manager which can be used by different kinds of plug-in "data handlers." Open source developers can write their own data handles for geographic data, for example. Oracle and DB2 also support such objects instead of just data in rows and tables. Red Hat's most recent version of its Red Hat Database has a highly automated installer. Again, the proprietary vendors also are automating their products.

But the open databases tend to be far simpler than their upscale rivals in all these areas. That's a major reason for their fast read performance, advocates say. And it also accounts for the low operational overhead.

"When you have a much less complex product, there are fewer things to watch, and fewer things to go wrong," Zawodny says. "You don't need a full-time [database administrator] looking after it. And that makes [the open database] cheaper, too."

To some, that same simplicity can make the open products seem "brain-dead" compared with Oracle and DB2's features. Stored procedures and triggers, which have been relational programming standards for more than a decade in commercial products, are missing from some open products, such as MySQL, but not from PostgreSQL. Both features will be added to MySQL this year, according to the company.

A common criticism is that open databases don't support transactions or don't do so as well as the commercial products.

"MySQL has a fast database for content store, but it is still immature in terms of transaction processing on the back end," says Charles Garry, an analyst for Gartner, assessing open databases for a Gartner teleconference on server infrastructure.

Advocates say transaction support in open databases is improving.

Grown-up software

Immaturity in some areas of an open database might not be a problem if the software has what you need in other areas, or has a credible track record of delivering new features on a regular basis.

The open source model is not monolithic: There are variations on a theme. PostgreSQL.org is emblematic of the free source code model, in which the database code is not owned by anyone, and the community organizes development and enhancements.

By contrast, MySQL was developed in the mid-1990s as a Web database by three Scandinavian programmers. The company freely offers the source code under the GNU GPL. But it also offers a commercial license. According to MySQL CEO Marten Mickos, Cisco is a commercial licensee of MySQL, incorporating it in five products, including its network intrusion-detection software. "They came to us and said, 'we love your product, but our lawyers won't let us use it under the GPL,'" Mickos says. With the commercial license, any changes Cisco engineers make to the source code remain Cisco's intellectual property and a distinguishing feature of its product.

Yahoo has deployed MySQL under the GPL license. "I can download it, change it, use it my way forever, and never pay them a dime," Zawodny says. Another benefit is the community that springs up around a successful open source program, fixing, refining and improving it.

Variety spices open source databases

Here’s the rundown on key open source databases.
Product MySQL 4.0 Red Hat database 2.1 SAP DB 7.4
LicensingGNU General Public License optional commercial license.No license specified.GNU General Public License.
Features

• Supports “broad subset” of ANSI SQL99 standard.

• Runs on Linux, Microsoft Windows, various Unix platforms.

• Modular design lets users plug in different storage engines to suit applications.

• Supports “core constructs” of ANSI SQL92 standard.

• Runs on Red Hat Linux 7.3, 8.0 and Advanced Server.

• Mature DB project pioneered 25 years ago at the University of California, Berkeley.

• Supports SQL92, entry level.

• Runs on Linux (Intel), Microsoft Windows, leading Unix brands.

• Designed for high-volume, online transaction processing; 100 SAP developers handle development.
Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2003 IDG Communications, Inc.