Data warehousing utopia

Dataupia, pronounced Datopia and standing for data utopia, has introduced its Satori Server 12000 as a data warehousing (DW) appliance with a difference. It combines industry standard processing and storage hardware and the Linux O/S with its own secret sauce, middleware to optimise business intelligence queries from standard databases and run them in a massively parallel processing system.

Think in terms of an HP Neoview system and not a much less processor-intensive business intelligence (BI) box like Sun's X4500.

The 12000 is a Lego block-like collection of blades. Samantha Stone, a Dataupia VP says it isn't a pre-configured appliance. That's true but it is composed of pre-configured components. That's how Dataupia manages the trick of offering an appliance that can grow.

A minimum system is a single blade. You simply add more blades to increase both storage capacity and processing power. The blades work together as a massively parallel system. They talk to the outside world via gigabit Ethernet.

Each blade houses a pair of dual 64-bit AMD Opteron CPUs. These get the data they work on from eight serial ATA (SATA) disk drives providing 2.2TB of storage capacity on the blade. The blade runs a version of Linux. So far, so industry standard. Now we get into the proprietary software.

A layer of middleware called Global Services is layered onto Linux and acts as the communicating and otimising engine. It co-ordinates the various blades and optimises BI queries to run on the 12000. It talks via a protocol engine to standard databases: DB2; Oracle; and SQL Server. Customers have no need to buy a separate DW database and continue to use their normal database from which data is extracted to load the Satori Server. It means they don't have to learn significant new database skills to run the Dataupia DW appliance. There the data resides in Dataupia's own database engine.

Dataupia has tested up to 50 blades working together. They are housed in a standard 19-inch rack unit by the way. Such a system would have 1.1PB of capacity and 100 CPUs.

Pricing is pretty straightforward, reflecting the simple building block architecture of the system. It is $19,000 (£10,000) per blade. Stone says: "It's everything you need to buy to upgrade your existing Oracle database. There is no initial high purchase cost."

The Dataupia idea is that by combining specialised software with blade building blocks composed of industry-standard CPUs, storage and Linux, it can offer a much more affordable and capacious BI system, enabling businesses to hold more data online and apply BI disciplines to a larger proportion of its growing data mountain.

It's notable that, from a storage point of view, direct-attach is the way to go for appliances. By definition they are self-contained and must have heir own hard drive storage; no storage area network or network-attached file system for them. This means that if the appliance approach spreads then direct-attached storage will spread a little too.

Dataupia was founded in 2005 by Foster Hinshaw, known as the father of DW appliances and the designer of the Netezza system, and John O'Brien.

Satori, a term from Zen Buddhism, means deep insight by the way.