ICL Plc is pulling out all the stops to convince us that it is a very different company from the one that failed all through the 1970s to capitalise on its CAFS Content-Addressable File Store breakthrough all through the 1970s, and one of its weapons will be the relational database engine for commercial transaction processing […]
ICL Plc is pulling out all the stops to convince us that it is a very different company from the one that failed all through the 1970s to capitalise on its CAFS Content-Addressable File Store breakthrough all through the 1970s, and one of its weapons will be the relational database engine for commercial transaction processing that will come with up to 256 superscalar Sparc RISC chips (CI No 2,051). Thoroughly commercial, the development focus, on relational database and transaction processing performance is not aimed at the supercomputer sector. The ICL machine, not expected to appear onthe market until 1994, will use up to 128 pairs of superscalar Sparc RISC chips – one CPU for the application, the other to drivethe network – each with 64Mb memory. ICL is currently evaluating Texas Instruments Inc’s SuperSparc – supplied by Texas for ICL’s latest DRS 6000 Unix boxes – and Cypress Semiconductor Corp’s rival HyperSparcimplementation for use in the new machine.Once database vendors offer products that adapt applications to parallel architectures, SQL queries from multiple sources will be able to take advantage of parallelism without the need for users to re-design their applications. That is the watershed, the company says. Oracle Corp and Sybase Inc have already set in motion plans for parallel-enabled versions of their respective database engines, and other players are expected to follow suit. ICL aims to be at the forefront of a new wave of parallel systems designed specifically for commercial transaction processing. Gartner Group expects the market to start to gain momentum from 1995 or 1996. ICL’s parent, Fujitsu Ltd, had no hand in the development effort, but is keeping tabs on its adopted child’s progress. Once the work is complete, the Japanese company could negotiate a marketing deal for the technology, although it is currently occupied by its own Sparc-based parallel processing effort, but that is for the scientific supercomputer market.