Startup Netezza Pushes Discount Data Warehouse ProductsStartup Netezza Pushes Discount Data Warehouse Products
The company is promising appliances that it says can tackle terabyte-scale business-intelligence apps 10 to 20 times faster than traditional systems at half the cost.
Building a data warehouse usually means buying and assembling general-purpose servers, storage systems, and database software--and spending millions of dollars in the process. This week, startup Netezza Corp. will debut a line of data warehouse appliances that it says can tackle terabyte-scale business-intelligence applications 10 to 20 times faster than traditional systems at half the cost.
The Performance Servers combine massively parallel storage with Hewlett-Packard ProLiant servers linked by Gigabit Ethernet connections and custom microprocessors that process data before sending it to the rack-mounted ProLiant CPUs. The custom chips, called "snippet-processing units," are critical to the server's speed. Netezza execs say they help the CPUs process data almost as quickly as it can be pulled from the disks, even processing data coming in from outside sources in real time.
Epsilon Data Management Inc., which sells data-management and database-marketing services, is trying out the Netezza servers. "The queries do fly," says Mike Coakley, an Epsilon marketing VP. The company is considering the replacement of its network of data marts with a single Netezza system for a direct-mail client.
Netezza's system runs on Red Hat Linux and a PostgreSQL open-source database. Companies use business-intelligence software from vendors such as Business Objects, Cognos, and MicroStrategy as front-end applications, as well as data-extraction and -transformation software from third-party vendors.
The Netezza product line consists of three models priced from $622,000 to $2.5 million, with raw storage capacity of 4.5 terabytes to 18 terabytes. All three are available now. Data warehouse systems across the industry average $2.5 million, while large-scale systems can cost $10 million or more.
About the Author
You May Also Like