FREE Subscription to Dr. Dobb’s Digest: Same Great Content, New Digital Edition
Site Archive (Complete)
Architecture & Design
Email
Print
Reprint

add to:
Del.icio.us
Digg
Google
Furl
Slashdot
Y! MyWeb
Blink
November 12, 2002

Free Enterprise

(Page 2 of 3)

Sourceforge.Net Goes Commercial

Despite the successes of open source databases, it would be a mistake to assume these products are right for every case. In some instances, even staunch open source supporters must turn to proprietary technologies when they need advanced capabilities. Take the case of SourceForge.net, which long ran its community site for open source development on PostgreSQL. According to director Patrick McGovern, the site recently chose to migrate its back end database to IBM DB2.

"I don't want to say anything bad about PostgreSQL," McGovern explains. "It's been a great database for us. Many, many companies use it, many sites use it, and it's something that we've been very happy with. But we decided to move to DB2 for a couple of different reasons. One is the growth of the site—we're always trying to catch up to the growth curve."

Since its launch, SourceForge.net has gained popularity with each passing month. The site currently hosts some fifty thousand open source projects, a figure that increases by roughly seventy new projects every day. And each day, around 700 new users sign on. Those figures translate into heavy traffic—as many as four million page views in a single day. McGovern is hoping that the connection pooling and replication capabilities of DB2 8.1 will help the site get over the growing pains it has experienced running PostgreSQL.

Jeff Jones, director of strategy for IBM's data management division, believes performance is one of the strongest arguments in favor of large commercial databases. "Our investment in DB2 goes to making it capable of running in clusters of hundreds of servers working on the problem simultaneously, or supporting terabytes of data. This is not territory where MySQL or PostgreSQL can go. So on the high end, there's no argument for open source databases. On the low end, there's a good one—but then, there are some functional limitations. The extensions that have grown into the larger competitors aren't present in the open source databases, in general."

Competitive Advantage

One of the strengths of open source is its community-driven development model. But according to Jones, the very fact that DB2 is a commercial product is one of the greatest contributing factors to its success. "The open source environment lacks one critical thing," he says, "and that's the urgency of being driven by customer requirements, with a professional development team that's measured by how quickly and how well they can meet those requirements."

Al Hilwa, product manager for Microsoft SQL Server, agrees. "Microsoft believes that the open source development model plays an important role in the software ecosystem. In fact, we believe that open source offers significant benefits. That said, I think we offer definite advantages with regard to long-term cost and value, choice of applications, services, and support. The fundamental differentiator between the 'big three' and the rest of the database contenders, cheap or free, is the ability to sustain the level of R&D to stay in the game as hardware, applications, and customer requirements evolve."

Those R&D dollars and the advanced features they yield are compelling reasons for many companies to side with the major commercial database vendors. As applications grow in complexity, the demands they place on the underlying database also increase. Truly enterprise-class systems are likely to outpace the abilities of open source software.

"[An enterprise-class application] is typically a clustered system," Woodward explains. "It has to provide read/write access to a data system at a level greater than what can be provided by a single system, [and] offers real-time redundancy, where a single machine can fail but leaves the total system operable. If your application is in this category, PostgreSQL is currently insufficient to meet your needs."

It's in these cases that the experience of major software vendors really shines. Jones explains that multi-system scalability has been a key focus for IBM's database development. "You really can't choose MySQL or PostgreSQL or SAP DB or any of the other open source databases if you want to get big," he says. "For smaller applications doing basic, standard, starter-set SQL, there isn't any reason not to choose MySQL or PostgreSQL. They're already there if you have a Linux box, and they may suit your needs. But there are things that you just can't do, in addition to scaling and performance, when you get big with these guys."

What Performance Level?

Before you discount open source databases on grounds of performance or scalability, however, it's important to keep your actual requirements in perspective. Independent developer Reuven Lerner feels the performance needs of many projects are overstated. "PostgreSQL is just fine when you have millions of rows. When you move into billions of rows of data, then it might make sense to consider Oracle. But if you have that much data, then you probably have sufficient money to cover the costs of a full-time DBA, high-end hardware, and...a full-blown Oracle installation."

Mark Woodward agrees that not every installation will benefit from an expensive commercial database. "The way I see it, there are three basic levels of SQL database deployments," he says. "The first level is the compact or simple level. This level is an excellent candidate for MySQL, as long as MySQL supports the type of queries you intend to perform. [The second level] is typified by no need for redundancy—periodic backups suffice—and the ability for a single machine to handle the expected load. The third level is the enterprise level."

Woodward believes that, based on their real-world requirements, most projects are good candidates for open source databases. "The second level is, in my experience, 90 to 99 percent of the overall SQL marketplace. As machines and disk storage have become faster and cheaper, this middle level has become more and more capable. PostgreSQL fits best in this category. There is a lot of competition with both commercial and open source software. The choice is based on SQL implementation, price, performance, reliability, and support."

As with the commercial competitors, the various open source databases on the market have varying assets. (See the sidebar, "Open Source Choices," for a comparison of several different packages.) No one product will be right for every application, but they all share at least one common strength. "In terms of price, [open source] wins hands down," says Woodward. "It's free."

Licensing and Support

Surprisingly, the restrictions of open source licenses are seldom a concern for database buyers. "If you dig carefully, most of the open source products are also available with different licensing models to comfort the buyer," admits Microsoft's Hilwa. For example, MySQL AB will license its product under either the GNU General Public License (GPL) or an alternate, commercial license.

As MySQL AB's Mickos explains, "The commercial license has nothing of the strict regulations of the GPL, so this is the way for commercial customers to get 'relief' from the GPL. Also, the GPL is an 'as-is' license, with no warranties or representations whatsoever. In our commercial license, we assume responsibility for the product in the same way as traditional software vendors."

This distinction is significant, as support remains a top concern for many potential open source adopters. According to Microsoft's Hilwa, "The market seems to demand a level of support, available skills, and deployment track record that imply a critical mass that many [open source] products do not have."

Still, it would be a mistake to assume that you're on your own once you've installed an open source database. Even those customers who license MySQL under the GPL can obtain support from a number of third-party vendors, and the same is true of the other open source competitors. In the case of PostgreSQL, building a solid support structure is a top priority among the development community. "PostgreSQL Inc. very recently got a global operation happening, so they can give global 24x7 support," says Justin Clift.

Though SourceForge.net's McGovern appreciates the support options his company's partnership with IBM makes available, he remains a believer in the community approach of open source. "We've got IBM we can call on and get questions answered. They've been very helpful with the migration [from PostgreSQL] up until now," he says. "But the open source community is often times as responsive."

Previous Page | 1 | 2 | 3 Next Page
TOP 5 ARTICLES
No Top Articles.



MICROSITES
FEATURED TOPIC

ADDITIONAL TOPICS

INFO-LINK