Oracle, IBM and Microsoft may have controlled almost 90% of the relational database market last year, but alternative products from vendors promising software that’s less costly and easier to manage are now appearing on IT radar screens.
Many of the upstart vendors are aiming for niches where conventional large database servers would be too slow or pricey. For example technologies such as handheld devices, point-of-sale systems and networking and security gear increasingly need to gather, store and process data in real time, a task that even the fastest Oracle or IBM DB2 database isn’t suited for.
That’s pushing some users towards embedded databases, such as an object-oriented one offered by Db4objects. The company’s open-source db4o software can be built directly into Java and .Net applications. Db4objects recently released Version 5.0 of the database, adding support for writing queries in the native semantics of Java, C# and Visual Basic .Net.
“Development with db4o is extremely intuitive,” says Matt Eskandar, technical manager at Canadian firm MR Control Systems, which sells industrial control software to power and oil companies.
MR, which is embedding db4o in a forthcoming version of its software, has used other databases in the past, including SQL Server 2000, Access, Paradox and dBase IV. “Storing an object using db4o takes only a few lines of code,” Eskandar says. “In comparison, putting together a parameterised SQL command to insert a record can become quite an exercise.”
Microsoft, Oracle and IBM all sell embedded versions of their databases that would compete with db40 but, despite limited marketing, Db4objects has gained more than 9,000 registered users in the past year, says its chief executive, Christoff Wittig. These include customers such as BMW, which is building a prototype Java-based control system for its luxury cars using db4o and Boeing, which is using the database in its P-8A marine aircraft for the US Navy, Wittig says.
Even some of the big database vendors are acknowledging the importance of technologies outside the relational mainstream. Oracle even put some of its money down on one last year, when it bought TimesTen, a developer of a database that processes information while it’s still in a server’s memory. Oracle released an update of the Times Ten software last month and is positioning the in-memory technology as a front-end cache for its relational databases.
Other vendors of database alternatives are also trying to meet users’ need for more speed than relational software can provide. For example, StreamBase Systems claims that its namesake database can process up to 150,000 messages per second as information flows through a system.
Another niche vendor, Ants Software, is trying to steal users away from the top vendors with technology it says can guarantee data integrity despite having to lock individual data cells only occasionally. Typically databases lock whole rows of data to ensure that information is written to each cell in the proper order.
Although analysts generally applaud the potential performance, manageability and pricing benefits of many of the emerging technologies, they also say the reality of user inertia may make it difficult for small vendors to win converts.
“Even if the technology does all that Ants claims, it’s going to be hard for them,” says Curt Monash, a consultant and US Computerworld columnist. “Most customers would rather just throw hardware at the problem.”