NEC to roll out new x86 fault-tolerant servers

By Joe Spurr, News Writer 16 Feb 2006 | SearchDataCenter.com

IT infrastructure news Digg This!     StumbleUpon     Del.icio.us

The Rancho Cordova, Calif.,-based company recently announced the Express5800/320Ma will cost its network a total of five minutes of downtime per year, with a choice of either dual-core 2.8 GHz Intel Xeon processors or single-core chips of 3.2 or 3.6 GHz.

More fault-tolerant servers: What's your downtime worth? HP NonStop server targets smaller businesses

Fault-tolerant systems have twin components that mirror, or lockstep, each other, so if one fails, the other continues to work with no interruption.

Extreme reliability, traditionally associated with larger systems, such as mainframes, has increasingly moved into x86 systems in the last few years as they are being pushed further up the IT food chain.

The x86 servers are being used in circumstances that are completely different from what they were used for to begin with, according to Charles King, principal analyst with Hayward, Calif.,-based Pund-IT Research.

"People are deploying [x86 servers] for a variety of business-critical applications," King said. "Which brings up the issue that if you are going to be deploying a low-to-midrange solution for high-liability applications, how do you ensure that you can depend on this to keep things running?"

NEC said the fault tolerant system holds an advantage over clustering by saving on license fees and labor cost.

"That's the secret sauce," said Dick Csaplar, a product manager for NEC. "Even though we have four CPUs in the box, we only need one copy of Windows. And all of the hardware is swappable. If you lose a module, you literally just slide it out of the server, ship it to NEC and we ship it back."

And by putting the processor and I/O onto the same module, NEC has reduced the number of modules from four to two. Less modules means the systems are less complicated and easier to maintain, Csaplar said.

The Express5800/320Ma will also feature Active Upgrade support as an option, which lets users upgrade without having to reboot by modifying one while the other handles the workload, then swapping again afterward to 'wash the other hand.'

Additional server rollback capabilities aim to protect data regarding such functions as security patches and service packs.

Tentatively available in March/April for $30,000-$50,000, the new systems will support Microsoft's Windows Server 2003 and Red Hat Linux sometime midyear.

Let us know what you think about the story; e-mail: Joe Spurr, News Writer

Tags: Data center disaster recovery planning,  Rack mount server and x86 hardware,  VIEW ALL TAGS

Digg This!     StumbleUpon     Del.icio.us