New Breed of Blade Servers Comes of Age

A new wave of blade servers from several top-tier companies will include attributes more commonly associated with full-size servers, such as streamlined manageability.

A new wave of blade servers from several top-tier companies will include attributes more commonly associated with full-size servers, such as streamlined manageability.

NEC Solutions America Inc., RLX Technologies Inc. and IBM are each readying new blade, or ultradense, servers that offer such high-end capabilities as fault tolerance, system management and storage, respectively.

The moves, along with the form factors existing attributes, such as modularity, make the systems even more desirable for some users.

"If we start adding a lot of subscribers now, we have to buy a whole new server," said Dan Carney, senior director of operations at Wildfire Communications Inc., which sells voice-activated communications products and services. "But with blades, all I have to do is buy another blade. The modularity of the blade is what Im looking forward to."

Wildfire, of Waltham, Mass., runs NECs existing Express5800/320La line of fault-tolerant, full-size servers but is considering the Boxboro, Mass., companys forthcoming blade, the Express5800/320Lb.

At the core of the systems fault tolerance capabilities is an identical backup of every component within the same blade. And unlike most blade systems, in which the blades stand vertically side by side, the 320Lb will use the traditional 1U (1.75-inch) horizontal form factor, according to NEC officials. Each chassis will house six blades, including two processor modules that will house one or two Intel Corp. 2.4GHz Xeon chips.

NEC rolled out the 320Lb blades in Japan last month and is launching them in the United States next month.

Like Wildfire, Persist Technologies Inc. is looking to blades to provide high availability to its applications. Using its own management software, Persist mirrors each of the one-processor RLX blades it uses so that if one fails, another can step up and perform the task, said Gary Lyng, vice president for product management and alliances for the Pleasanton, Calif., company. The failed blade can then be replaced without interrupting the service.

"We develop a blade-box environment without one single point of failure and [with] fault tolerance," Lyng said.

RLX, of The Woodlands, Texas, is keying off the concept of ease of management with a new overarching strategy called ActiveIT, which comprises blades, tools for clustering, blade management and provisioning, and professional services.

RLXs new dual-processor ServerBlade 2800i/3000i is equipped with Intel Xeon processors that run at 2.8GHz and 3GHz. It can run either Red Hat Inc.s Red Hat Linux 7.3 or Microsoft Corp.s Windows 2000. A new 6U (10.5-inch) chassis can hold up to 10 blades. Each blade also features an LCD that indicates whether there is a problem.

In addition, IBM, of Armonk, N.Y., last week announced it is extending the reach of its BladeCenter servers by offering embedded storage area network and Fibre Channel connectivity in them. The move will ease the physical management of the boxes by reducing the amount of cabling required.

Component makers are also beginning to create products specifically for blades. Intel, at its Intel Developer Forum in San Jose, Calif., this week, will give details on Deerfield, its 64-bit Itanium processor for blades.

Sources familiar with the Santa Clara, Calif., companys plans said Deerfield will consume at least half of the 130 watts of power that the current Itanium 2 chip uses. Deerfield is due in the second half of the year.