LinuxWorld

A faster, sleeker way to do MySQL?

Start-up offers appliance delivering high-performance data warehousing

The archetypal MySQL database chugs along in the server room of a Web 2.0 start-up, powering a small but promising Web site. In fact, a quarter of the 12 million MySQL installations are data warehouses doing business intelligence and analytics for companies.

Slideshow: Five fantastic open source tools for Windows admins

The problem is that without labor-intensive retuning, MySQL performs poorly as a data warehouse.

"If you've got a lot of time on your hands, you could mess around with MySQL," said James Kobielus, an analyst at Forrester Research Inc. Otherwise, "it's just not optimized for BI."

Today, a start-up is officially unveiling a MySQL-based appliance boasting proprietary chip and software technology that it claims delivers high-performance data warehousing out of the box.

Santa Clara, Calif.-based Kickfire Inc. is starting to ship its appliance in earnest this week. Kickfire is targeting what it says is the mass market for MySQL data warehousing -- smaller-than-10TB warehouses used for Web analytics.

Based on a SQL chip developed by Kickfire's founders and running on hardware from Sun Microsystems Inc., the Kickfire appliance starts at $32,000 with 1TB of storage.

Karl Van den Bergh, Kickfire's vice president of marketing, admits that the figure is about twice the price of a general-purpose server running the open-source database.

However, he also claims that Kickfire more than makes up for that price by delivering far faster performance than typical MySQL servers. He credits its SQL chip, which accelerates the database by running commands in silicon, as well as its columnar storage engine tuned to read data quickly, which is key for data warehouses.

Mamasource is a five-year-old Web start-up in Corte Madera, Calif., that is Kickfire's first named customer. (Kickfire has also claimed several unnamed customers in the past.) The company, which operates an online community for moms, was running a 300GB MySQL data warehouse to analyze its user clickstream data. Performance was sluggish. For instance, it took about four hours a day to upload 1GB of clickstream data into the data warehouse, during which time the data warehouse was fully occupied, according to Steve Ketchpel, director of research and analytics at Mamasource.


Invalid query - session: Can't connect to local MySQL server through socket '/tmp/mysql.sock' (2)
Newsletter sign-up

Sign up for one of Network World's newsletters compliments of Linux World

Linux & Open Source News Alert
Web Applications Alert
Video and Podcast Alert
Security Alert
Virtualization Alert

Email Address: