Energised by its agreement to acquire PowerSoft Corp a few weeks back, and keen to take a swipe at finger-pointers casting technological aspersions in its general direction, Sybase Inc, as reported briefly (CI No 2,556), has unveiled a database query tool, IQ Accelerator, which it claims can conduct queries across different databases regardless of the hardware or software storing the data, and produce almost instant answers. Sybase picked up IQ Accelerator with its acquisition of Expressway Technologies Inc of Waltham, Massachusetts, back in October (CI No 2,535) and says query responses are up to 516 times faster than queries on SQL Server alone, and up to 240 times faster than queries on DB2 running on the mainframe. As Expressway III, the technology has been testing at 10 sites since April. Details of the Bit-Wise mechanism IQ Accelerator uses are sketchy; Expressway is seeking a patent for the technology. However, the technique enables all kinds of data, including characters and numeric fields, to be represented as bits in a bit-map, circumventing the need for table scans using conventional 'B-Tree' balanced tree indexing which Sybase claims is slower and takes up more space. Compression enables data to be stored in the index at 50% to 75% of its inflated size. An integrated Sybase SQL Server release incorporating IQ Accelerator will test in the spring and is generally available next summer, initially for AT&T Corp, Data General Inc, Hewlett-Packard Co, IBM Corp and Sun Microsystems Inc Unix boxes and Windows NT. IQ Accelerator, which accepts queries in an ad hoc or unstructured way, operates on Oracle Corp and Informix Software Inc databases as well as Sybase and IBM Corp ones, though the firm says it has not yet decided whether to market packaged IQ Accelerator for rival relational offerings.
But Sybase is planning to use IQ Accelerator in its search for other revenue pastures and has decided to throw itself at the market for data warehousing, a concept regarded by sceptics as the computing equivalent of the brick privy at the bottom of the garden but described by analysts as a burgeoning $750m market growing at some 40% a year. Sybase is offering a grab bag of what it describes as data warehousing 'glue' in the form of Warehouse Works, a marketing programme built on its Open Client and Open Server Application Programming Interfaces. The company is now offering all of its technologies - Sybase SQL Server, Replication Server, Enterprise Connect and IQ Accelerator - as viable components of a data warehouse with some 125 tools from third parties expected to fill in the gaps. Sybase, which lacks integrated front-end application tools, will work with existing partners such as Prism Solutions Inc to provide technology in four key data warehousing areas it identifies as data assembly, transformation, distribution and access. In addition Replication Server is being enhanced to support replication from non-Sybase databases - both mainframe and other local network databases. In launching itself directly against established data warehouse suppliers such as Red Brick Systems Inc, Sybase has - at least according to one player - become the first mainstream company to acknowledge the need for separate, specialised offerings to address the transaction and data warehouse markets. Sybase claims its products can be used to create data warehouse databases that are typically smaller than the sum of the component data they contain, whereas other suppliers typically require fully-indexed data warehouses to be larger than the data pool itself. Moreover, IQ Accelerator does not require the creation of time and space-consuming 'pre-tables' a la Red Brick, the company says. Data warehousing is essentially the ability to apply complex and ad hoc queries to vast amounts of historical data which may have been spread all around an organisation, in pursuit of underlying trends and marketing information such as identifying customer prospects and subsequently ascertaining whether those prospects are typically cash or credit spenders.
A data warehouse database differs from a traditional transactional, relational database by storing data over time rather than regularly updating information. Data warehousing requires lots of input-output operations to respond to typical queries - the problem is getting the data out: transaction systems demand sub-second responses to queries that usually require just a handful of input-output operations. But they argue that most of the technological issues in data warehousing have been solved and that the question now is one of what combination?