Currently being referred to as Oracle10i, but unlikely to ship under that name, according to Oracle’s chief marketing officer Mark Jarvis, the next version of Redwood Shores, California-based Oracle’s database is due to be launched in September 2003.
Details of the new version are currently thin on the ground, but Jarvis revealed that the focus of development is Linux clustering and grid technologies. 10i is being built to support heavy Linux clusters in a grid environment, he said. In 10i you’ll see a lot of features that will totally exploit the features of Linux and you will not get those features on other platforms.
Talking to ComputerWire, Jarvis refined that statement to explain that Oracle is building technologies into the database that Linux will be better positioned to exploit than other operating systems.
Linux is clearly the dominant growth operating systems and our goal is to leverage that growth and leverage clustering of Linux. The other major driver is grid computing, he said. A lot of the database development we have going on is designed to hit that sweet-spot.
Jarvis said that Oracle is developing technologies to enable the time-sharing of applications, resource on demand, and the ability to upgrade and add new systems to cluster grids without shutting them down. Like the Cluster File System – another recent addition to the Oracle database – these technologies will at first be available within Oracle, but will later be released to the open source community.
When we built CFS we built it and released it as an open source product, said Jarvis. It was built for open source, but was released on other platforms later. He said that although these features will be available for Oracle on all supported operating systems, they will be better exploited on Linux, which does not include competing functionality.
As an example, Jarvis pointed to the complications involved in running CFS on Microsoft Corp’s Windows, which has its own Cluster Service technologies. CFS is a part of the database on Windows, but it’s a part of the operating system on Linux, he said.
CFS is designed to simplify the management of Oracle9i Real Application Clusters, but according to Jarvis more cluster management technologies are needed for Linux clusters. When you start to cluster Intel Linux systems you need to manage those systems, and the management tools for that are largely non-existent today, he said.
Jarvis said he was referring to the management of database clusters, rather than management of application clusters, for which there are a multitude of open source and commercial tools available.
The [database cluster] management tools today are good at managing individual servers, but not at treating the cluster as a single system, he said. Jarvis added that what needs some work are lock-management tools that can handle the locking between cluster nodes to ensure that database processing tasks are not repeated across the cluster.
Jarvis confirmed that Oracle will continue to contribute database technologies to the open source community, but ruled out the potential appearance of an open source version of the Oracle database.
There’s a bunch of stuff being built that we will contribute to the open source community because it will be of benefit to all Linux vendors. The biggest so far is CFS, but a lot of similar pieces will come, he said. When you build a database there are logical areas to interface to Linux that can be open sourced, but we will never open source the core database.
Source: Computerwire
