Data warehouse (DW) startup company Kickfire announced its new query accelerator's beta release in April. The Kickfire Database Appliance has a built-in SQL chip, and pulls data directly from memory, bypassing registers and cache, loosening I/O bottlenecks. General availability is scheduled for Fall 2008. The device breaks an SQL query into 'parallel execution plans' which are then fed to the chip. In parallel, the searched-for data is retrieved from memory in compressed format, and processed as it passes through the SQL chip. Kickfire's marketing VP stated that the SQL chip fits into the hardware and connects exactly the same way to the host server as a graphic chip does. And the SQL chip co-processes with the general-purpose host server in the same way. However, built-in parallelism guarantees the same load speeds provided by multiple CPUs combined. Incremental load capabilities for MySQL provide that changes on the source database will automatically move into the Kickfire Database Appliance. Although the DW space is currently dominated by mammoth companies such as IBM, Oracle and Teradata, open-source database companies are making market inroads with, for example, PostgreSQL-based Greenplum and MySQL-compatible Infobright. The research director of The Data Warehousing Institute said the Kickfire Database Appliance could also accelerate DW market adoption of MySQL technology, which actually needed the fine tuning for query processing and data warehousing provided by Kickfire's device. Kickfire joins 'a rash of startups in the past 18 months' trying to improve DW query performance. The company decided on MySQL for technical as well as business reasons. It supports pluggable storage engines that use MySQL standards, so that Kickfire could create its innovations with the SQL chip and database kernel underneath the chip, and still be part of a standard database.