The first thing Geoff Squire did when he was appointed as chairman of UK data warehousing systems hopeful WhiteCross Systems Ltd (CI No 3,039) was to show three of the company’s four original investors the door. Squire, the former head of worldwide field operations at database company Oracle Corp and chief executive of systems management […]
The first thing Geoff Squire did when he was appointed as chairman of UK data warehousing systems hopeful WhiteCross Systems Ltd (CI No 3,039) was to show three of the company’s four original investors the door. Squire, the former head of worldwide field operations at database company Oracle Corp and chief executive of systems management software company OpenVision Inc, currently in the process of merging with Veritas Software Corp, had been drafted in to build a company out of a start-up that had already started to flag. WhiteCross sprang to life in 1992 with the arrival of a group of European sales and engineering refugees from Teradata Corp, the pioneer in data warehousing systems, that had then just been acquired by NCR Corp. Led by Dan Holle, the original developer of the Teradata machine’s all-important query optimiser, WhiteCross set about building a machine where the sheer brute force of the processing engine would allow data analysts to retrieve answers from huge volumes of data in mere seconds. It sold some early systems starting in 1994 – almost all in the UK – but without enough venture capital to build up sales and marketing globally, WhiteCross was swimming upstream, according to US-born Holle. Squire went looking for a new management team and a fresh injection of venture capital. He appointed a veteran manager with plenty of experience of selling UK-crafted products into the US: Chris Barfield, former head of project management software company Metier Inc. The money, around $6.5m, came from venture capitalist Warburg Pincus. Those changes coupled with the fact that customers are reporting back that the company has now got the machine running and scaling well, gives WhiteCross confidence that it can carve a niche out of the booming data warehousing market. Annual revenues at this point are still small – probably around $6m. Last year the company won three deals over $1m for 100 processor machines, bringing its installed base to 30 at 20 customer sites, including British Telecom, Scottish Power, Mercury, and the London Future and Options Exchange. They are using WhiteCross WX9010/20 systems, massively parallel processing computers, in configurations that harness from 12 to 204 Transputer processors. The company has also just sold its first system into the US – a 204-processor machine to Chrysler Finance Corp which will use it to analyse the credit-worthiness of leasing clients – after the appointment of a US chief executive officer, Jim Suszka, ex-Metier Inc. In the bidding for the Chrysler deal, which represents WhiteCross’ largest contract to date, a WhiteCross system was up against an IBM SP2 massively parallel processor. WhiteCross ran all of the 4,000 queries that Chrysler staff most commonly run in a couple of hours; the IBM machine took twenty minutes to run the first query. The WhiteCross sales message is quite simple. It says that many data warehousing projects have failed in their goal of providing free-ranging data access and analysis facilities to line managers. Where end-users can get at the data, they can only do so if the data is pre-indexed or summarized, or the queries are pre-defined. By throwing processing horsepower at the problem and splitting the query over hundreds of processors, problems can be cracked within seconds rather than hours. The payback on the $1m plus WhiteCross systems comes in a matter of weeks, says Barfield. And the company is willing to offer guarantees. In a marketing campaign, it states we guarantee that you will realise a positive payback from an agreed data exploration project or the fee will be refunded. That is bold talk for a company at such an early stage in the product lifecycle. WhiteCross is also bucking a trend. In recent years, companies building high-end data querying engines have switched away from providing a dedicated database machine. NCR’s Teradata product, the inspiration for the Whitecross system, is now implemented under Unix on a general purpose NCR server. Red Brick Systems Inc also offers a competitive product that runs on standard hardware. And, historically, other attempts at a data-querying machine have failed – ShareBase Inc and Charles River Data Systems Inc disappeared from view in the early 1990s. WhiteCross argues that the need for superfast querying will overcome any nervousness potential customers have about the technology – and the guarantee will back that up.