HP Gives Scale-Out Architecture Extreme Makeover

HP on Wednesday announced its Extreme Scale-Out (ExSO) portfolio, intended to cut data center costs for businesses involved in heavy Web 2.0, cloud computing and high-performance computing activities.

Such companies typically have data centers with thousands of servers.

The HP ExSO portfolio includes a lightweight modular system architecture, as well as services and support.

However, HP is in some ways already behind archrival IBM, which last year unveiled its System x iDataPlex solution, for the same target markets.

In fact, both may be missing parts of the market; many companies with huge data center demands buy cheap components directly from manufacturers and build their own servers anyway.

Details of ExSO

HP’s scale-out portfolio currently includes the HP ProLiant DL1000 Multi-Node servers unveiled last week; the HP Performance Optimized Datacenter (HP POD); the HP StorageWorks 9100 Extreme Data Storage System; and the HP ProLiant 2x220c double-density blade server, introduced last year.

ExSO is built around the ProLiant SL 6000 server family, which replaces the traditional chassis and racks with a lightweight tray and rail design. It’s like swapping out a chest of drawers for an open-front bookcase.

Increasing demand led HP to develop its ExSO portfolio. “We’ve had a scale-out business for many years,” Ed Turkel, manager of scalable computing and infrastructure business development at HP, told TechNewsWorld.

“Now the scale-out computing market is growing with new companies in Web 2.0 and [the] cloud, and we’re seeing increased interest from financial services, oil and gas, healthcare, as well as media and entertainment companies,” he said.

Keeping an Eye on Data Centers

The lightweight new architecture reduces costs, as well as the amount of data center space required, HP said.

Data centers are monitored with another part of the ExSO solution, the HP Data Center Environmental Edge. This uses small sensors to monitor power and cooling distribution around the clock every day of the year. It lets administrators perform root cause analysis to better manage costs.

ExSO also comes with a variety of services in the HP Services for Scalable Computing offering. These include technical account management installation services; value-added services; and on-site hardware support.

About the ProLiant SL 6000

The ProLiant SL 6000 family of servers is designed around extreme scale-out, high-performance computing and is aimed at data centers with power and space limitations. It consists of pluggable server modules in a standard 19-inch rack.

The servers have a high-efficiency 2U power and cooling chassis. A rack unit, or U, is 1.75 inches high; “2U” refers to two rack units.

They come with a choice of power supply: 60W; 750W; or 1,200W AC.

All the servers come with HP embedded SATA controllers with RAID capability.

The Competition

As is often the case, HP faces competition in this space from archrival Big Blue. IBM announced a similar system, its System x iDataPlex, last year.

Like HP’s ExSO, IBM’s system is designed to use less space and power. It has a rear-door heat exchanger that lets the system run at room temperature, reducing air conditioning requirements and, hence, power costs.

Last week, Dell made a scale-out data center announcement as well, Nick Simpson, senior analyst at the Burton Group, told TechNewsWorld.

“This sort of thing is for data centers, where scaling is done with lots of small systems and you aren’t looking for servers to run really big loads,” Simpson said.

DIY: Cheaper and Bigger?

The really big boys, however — the Googles, Yahoos and Facebooks of the world — often buy cheap components directly from manufacturers in the Asia-Pacific regions and put their own systems together.

“When you pay Dell, HP or IBM, you pay for equipment and components, as well as for the brand, the support and systems,” Richard Jones, vice president and service director at the Burton Group, told TechNewsWorld.

“But people building very large data centers don’t care for the support,” he said. “When one of their servers goes bad, they just pull it out and throw it away.”

Leave a Comment

Please sign in to post or reply to a comment. New users create a free account.

More by Richard Adhikari
More in Data Management

LinuxInsider Channels