Oracle has unveiled the first major version of the TimesTen in-memory relational database developed in house since acquiring the technology in June 2005, emphasizing increased integration and caching with its enterprise-level Oracle Database 10g.

The vendor made Oracle TimesTen In-Memory Database Release 7 generally available Tuesday. The previous version of the database, Release 6, which appeared in late September 2005, was almost in beta testing when Oracle purchased TimesTen, according to Jim Groff, senior vice president at Oracle, and the former CEO of TimesTen.

One main use of TimesTen is as a front-end data cache for Oracle 10g, Groff said. The TimesTen software can store a subset of the vast amount of data held in the back-end 10g database in memory as part of a company's middleware layer, where it can be accessed more quickly than querying the entire enterprise database. The rapid access to information can significantly boost the response times of applications.

All the new functionality in TimesTen 7 is related to Oracle 10g, but the vendor could offer the same capabilities to support third-party databases, Groff said. Some TimesTen customers, particularly on Wall Street, use rival databases from Sybase and IBM to connect to the in-memory database.

Along with real-time caching, TimesTen 7 includes the ability to handle 10g data types and SQL (Structured Query Language) features similar to 10g. The new version is also integrated with Oracle's Real Application Clusters (RAC) technology. If a network connection is broken, application processing can still continue in the cache and, once that connection is restored, the integration between TimesTen and RAC will enable automatic synchronization between the front-end and back-end databases.

TimesTen 7 also takes advantage of all the work Oracle's done on internationalizing its other software so that the new release supports over 50 database character sets and 80 languages. The database also supports Oracle's Fusion middleware and its SQL Developer and JDeveloper tools.

Since buying TimesTen, Oracle has been working to widen its usage, which was previously focused on financial services, telecommunications and the defense and intelligence sectors. The database is gaining customers in the retail, transportation and logistics industries, Groff said.

Oracle's main applications rival, SAP, has been talking up the benefits of using in-memory technologies to carry out search and business intelligence (BI). Oracle may look to make more use of the TimesTen technology in those areas, but Groff cautions that not all so-called "in-memory" functionalities are the same. "You see SAP making noise about in-memory, but we've been doing it since 1998," he said.

SAP debuted its BI Accelerator analytical engine in May and is readying its Enterprise Search software to appear in the first half of this year, both of which draw on in-memory capabilities. Oracle's rival Secure Enterprise Search shipped in the middle of last year and is based on Oracle 10g.

Looking ahead to TimesTen 8, "We have some work to do to bring everything under a common management framework," Groff said. Customers with hundreds of TimesTen caches accessing multiple Oracle 10g databases are keen to be able to manage the whole entity as a single integrated unit, he added.

Pricing for TimesTen 7 begins from $12,000 per CPU (central processing unit) and depends on the size of the in-memory database.