Many enterprise software users and vendors have made significant commitments to open source technologies. Projects such as the Linux operating system, Apache Web server, and the Perl programming language, have proven themselves as viable alternatives to equivalent commercial offerings. But what about the tools used to hold core information assets? Are open source databases ready for the enterprise?
Until recently, databases have been a relative unknown amongst open source software. However, the increasing adoption of open source database technology by well-known multinational organisations has led to more enterprise recognition and hence consideration.
There are many open source databases with different features of which MySQL and PostgreSQL are the best known.
According to the MySQL project, MySQL is the world’s most popular open source database server with over four million installations. From the MySQL Web site: “Customers such as Yahoo! Finance, MP3.com, Motorola, NASA, Silicon Graphics, and Texas Instruments use the MySQL server in mission-critical applications.”
However, “mission-critical” is vague when the exact details of the implementation are not revealed. Nevertheless, if the MySQL project can claim such high-profile users then the broader enterprise has reason to consider the technology.
“Absolutely!” Marten Mickos, CEO of MySQL AB, said when asked if he would recommend the use of an open source database in the enterprise.
“There are four main reasons for choosing an open source database,” Mickos told Computerworld. “To get better performance, better uptime and reliability, better ease of deployment, use and maintenance, and to get better economy.”
According to Mickos, the place for open source databases is rapidly expanding within the enterprise.
“Right now, open source DBMSs lend themselves to mission-critical heavy-duty Web, data warehousing, departmental, and use on the edge of the enterprise,” he said. “Increasingly, open source DBMSs power central business systems.”
When asked about some of the limitations of open source databases which may be preventing open source adoption, Mickos said: “Typically, open source DBMSs are superb when it comes to the core server technology, but they may still have some way to go when it comes to add-on products, third-party integration, and extensive professional services.”
MySQL AB offers support for the MySQL database as well as a commercial licence for the software.
Mickos believes although there are a number of large enterprises that depend on open source databases today, over the next year or two we will see a strong adoption of open source DBMSs in the enterprise.
PostgreSQL is regarded as a more mature and functional opposite number to MySQL in the open source database arena. Since its inception in 1986, PostgreSQL has features typically associated with commercial databases – such as transaction support, scalability and replication – and is well known for its stability.
“Once you get it up and running it’s great,” Marty Gauvin, managing director of Hostworks, said when asked about the company’s use of PostgreSQL. “Tuning PostgreSQL is more difficult than Oracle and there is less information on running high-end systems with it, however, PostgreSQL does have a level of production stability.”
Hostworks has a heterogeneous server environment consisting of Windows, Linux and Solaris. Two high-profile Hostworks clients, AOL|7 and Austereo, are running on PostgreSQL and MySQL backends respectively.
“These are media driven environments and don’t have much in the way of static database requirements,” Gauvin said. “Such sites have a large emphasis on throughput and the [open source] databases are able to deliver hundreds of transactions per second. Quite a lot of the traffic – such as interactivity and logging – is two-way, making read and write performance critical.”
As a user of commercial databases Gauvin commented on some of the differences between the two types.
“Oracle is more mature than MySQL, yet MySQL is maturing faster than Oracle,” he said. “The advantage of using Oracle and other commercial databases is the amount of skills available. For example, if I advertised for an Oracle administrator I would get plenty of applications.”
Regarding total cost of ownership, Gauvin said it’s really dependent on what you want to do with the database.
“If you’re doing weird things then the TCO of an open source database is likely to be higher whereas if you are doing something straightforward, the TCO should be lower,” he said. “However, the line is moving all the time.”
Although stating that open source databases are very reliable, Gauvin said the time taken to fix a problem is likely to be longer with an open source database.
What the vendors are saying
Although not widely deployed in the enterprise, open source databases have nonetheless attracted the attention of the tier-one vendors. Peter Fletcher, director of Sybase Australia and New Zealand, said the real issues with open source databases are availability and support, and not the technology.
“Our customers are from the government, health, financial, and telecommunications sectors. Their basic systems just can’t be down,” Fletcher said. “There are questions surrounding the ability of open source databases to support customers 24x7.”
Fletcher said open source databases are “free”, but from a support perspective they are not.
“In this respect, open source databases may or may not be appropriate for the enterprise,” he said. “Our customers are running very large systems so in the short term we see no threat to our business. Open source databases will only become a threat when there is a ‘buffer’ in the form of testing and support between the development community and the enterprise user.”
This sentiment is shared by Oracle, who as a company, does not view open source databases as appropriate for the enterprise.
“We do not believe open source databases are ready for any level of business,” Roland Slee, director of business and technology solutions, Oracle Australia, said. “The database captures the value of work and is the most critical software layer. If data is lost the organisation is put at risk.”
That said, Slee did express Oracle’s support for the open source movement.
“Open source has proven itself to be a viable operating system development environment,” he said. “Oracle has embraced Linux as an efficient, low-cost solution which we have given the ability to cluster industry-standard servers for mission-critical applications.”
Slee said the database will be the last bastion of commercial software due to its sophistication.
With major vendors not being overly concerned at the impact open source databases may have on their business, SAP is bucking the status quo with its own open source database – SAP DB.
Although having widely adopted Linux and other open source software for some time now, IBM stands by its flagship database, DB2, due to its feature set.
"Open source database software lacks some critical enterprise capabilities including scalability to support thousands of concurrent users and terabytes or petabytes of data," Brett Vincent, regional manager, DB2 information management software, IBM Australia and New Zealand.
"Many of DB2's more recent enhancements are also not seen in open source database software: heterogeneous information integration, native XML support, integrated business intelligence capabilities, and cost-based automatic query optimisation."
When asked whether IBM is likely to adopt an open source database in a manner similar to the company's Linux interests, Vincent said: "We continue to invest heavily in its [DB2's] continued growth and expansion. Speaking from the perspective of IBM DB2 information management software, this will remain our direction."
“SAP DB remains the only enterprise-class open source database and it competes well with Oracle and DB2,” Phillip Merrick, solutions architect, SAP Australia and New Zealand, said. “We open sourced SAP DB in order to give our customers better value and the overall reaction has been positive. The value for money has not changed a lot in the database space and we hope to change that.”
According to SAP, there are some 700 SAP DB users including large enterprises such as Colgate Palmolive, Toyota and Deutsche Post.
“SAP customers are happy with the reduction in capex and the support costs are not likely to increase,” Merrick said. “One of the benefits of using an open source database is the users are not forced into a perpetual upgrade cycle.”
The consensus with open source databases seems to be one of initial investment and support. If your enterprise is prepared to adopt the in-house support requirements of an open source database, then the free open source databases can save you from purchasing a commercial one. However, if your requirements involve high levels of support, a commercial database may be the most suitable option.
|The major open source databases|
|MySQL||1995||Free||GPL or commercial||www.mysql.com|
|SAP DB||2000 (OS)||Free||GPL/LGPL||www.sapdb.org|
|Berkeley DB||1991||Free||Open Source||www.sleepycat.com|