Original URL: http://www.theregister.co.uk/2008/04/27/kognitio_daas/
And now it's time for Data Warehouses as a Service
Kognitio monitors toaster inventory over the wire
Poor old Teradata - the data warehouse kingpin and lord over real-time global toaster inventory information. First it gets attacked by a host of data warehouse appliance makers, trying to undercut it on price. And now it's being hit by buzzword-based business models from those firing up DaaS (data warehouse as a service) systems.
Earlier this week, Teradata moved to outflank the appliance crowd by expanding its product line to include low-end and mid-range data warehouse gear. Just minutes after our story on Teradata's fresh kit ran, an e-mail arrived from Kognitio. Forget appliances, went the pitch. DaaS - or is it DWaaS - is where you'll find the real data warehouse action.
The technology roots behind Kognitio's play stretch back to a system running on a Thomson transputer and the real-time LynxOS operating system. Over the years, this approach has morphed into one centered on running the WX2 analytical database on top of Linux and x86 servers. Kognitio is also in the process of firming up Solaris x86 support.
Kognitio will sell customers WX2 on its own or help them create their own data warehousing appliances using the software in conjunction with the customers' own servers. And, late last year, Kognitio began a DaaS program, so that customers could tap into its data centers and pay for space on a large-scale data warehouse.
On the surface, the idea of a data warehouse service seems hard to fathom, especially for larger customers. You're sending serious volumes of data over the wire or via the mail for frequent analysis and then waiting to get the results back. This is perhaps a tough sell in a business that thrives on using sophisticated data analytics to make rapid decisions.
Kognitio, however, says it can handle the task at hand and points to BT as a giant customer that's already using the DaaS service in the UK.
"For BT, we work with their pricing department," John Thompson, Kognitio's US operations chief, told us. "They collect competitive tariff information and provide that to us. We re-price all BT's calls based on competitive tariffs."
The company brags about 14 minute load times for large volumes of data and 6 minutes for data transformations. So, it's the speedy attributes of WX2 that help it make sense under the DaaS model, according to Thompson.
"WX2 reduces upfront data management, modeling and load times dramatically," Thompson said. "With most merchant databases, you still have to do a lot of modeling, indexing and aggregating. With WX2, you don't have to do any of that."
Those of you in the UK might be familiar Whitecross Systems, whose IP Kognitio acquired in 2005 in order to keep developing WX2. Now, the company looks to heighten its presence in the US by selling WX2 and by creating a DaaS operation. Kognitio plans to have a US data center up and running in the next six weeks, Thompson said.
This is the part of the story where we usually tell you how Kognitio prices its wares. But Thompson declined to provide any specific information at this time, saying that pricing will fluctuate quite a bit based on the job. He would only say that the DaaS service comes in at half the cost of data warehouse appliances from competitors. If you're like us, then it's difficult to judge the veracity of such a line without, er, data to back it up.
Some appliances such as those from Netezza run on specialized hardware, while others from the likes of DATAllegro rely on specific stacks of partner hardware. Kognitio thinks it has the better approach whether you're doing DaaS or in-house work, since it will run on just about any x86 servers and storage.
"We submit SQL into machine code where it's running on x86 chips in native mode," Thompson said. "The reason people have to certify on certain platforms is because they're running SQL in an emulated mode."
Kognitio targets mid-sized to large businesses and is looking for companies in a wide variety of industries to embrace data warehousing. It's betting that many of these customers - be they life sciences or media firms - will turn to the DaaS model rather than dealing with these rather complex systems in-house.
You can find more information here. ®