ClearCube Technology plans to launch two PC blade servers, in a bid to compete with Hewlett-Packard Co. for customers in finance, insurance, hospitals and the military.
Both blades will use Intel Corp. chips and will have enough power to run Microsoft Corp.’s pending Vista operating system, which has large processing requirements to support the translucent windows in its graphical interface.
The R1300 blade will use a dual-core Pentium 4 chip, while the R2200 uses a pair of Xeon chips based on Intel’s Irwindale design. ClearCube plans to upgrade the R2200 to Intel’s 65-nanometer Woodcrest design by the fourth quarter.
“We wait about a quarter to use the latest processors, because customers like to see the price point drop and a three- to six-month track record,” said Raj Shah, chief marketing officer of ClearCube, in Austin, Texas.
ClearCube’s customers use the servers as shared resources, accessing them through simple desktop “user ports” that host almost no processing power or storage.
That model allows IT administrators to keep tight control over the security and upgrades of a company’s computers, since the valuable chips and memory remain in locked computer closets, not on employees’ desktops or laptops.
Both the R1300 and R2200 are backward compatible with previous ClearCube products, so they will work with infrastructure including the chassis around the blade, the backside connector modules and the user ports on desktops.
That continuity of design also makes it easier to reallocate older resources to employees with less compute-intensive jobs, in a “waterfall” effect.
“If you had bought the 2.2GHz version two years ago, you would get a 3.8GHz machine now for your trader or engineer or military analyst who needs the latest speeds and feeds. But a medical administrator could hold onto his server for three or four years,” Shah said.
ClearCube users increasingly devote their new servers to shared workload models like virtualization, which allows as many as 10 different users to share a single blade, and like grid computing, which divides large computational loads into smaller chunks for many different computers to handle in parallel.
Whatever model they choose, shared computing is a better approach for business than clusters of individual PCs, Shah said.
“The PC was not designed to be a business computer; it was designed to be a personal computer. We just all ended up doing it this way,” he said.
“Twenty years from now, if someone tried to sell you something called a ‘box PC’ that sat on your desk and had less uptime and less security, would you buy it? Of course not.”
Both the R1300 and R2200 blade PC models start at US$1,799.