Xyratex International
HPCwire

Since 1986 - Covering the Fastest Computers
in the World and the People Who Run Them

Language Flags

Visit additional Tabor Communication Publications

Datanami
Digital Manufacturing Report
HPC in the Cloud

HP's Double-Wide Assault on Scale Out Computing


Today HP merged its formerly low profile Scalable Datacenter Infrastructure (SDI) team -- a group within HP that has spent the last two years focused on 10-12 very large media and Web 2.0 companies -- with its HPC team, and announced a new scale out blade product that lets customers cram over 12 TFLOPS in a 42U rack by putting two dual-socket servers on a single blade. The newly formed Scalable Computing & Infrastructure group is positioned at the junction of traditional HPC and enterprise computing, a position that HP's Ed Turkel believes will allow the company to capture this growing segment of the market.

HP's newly announced Scalable Computing & Infrastructure group is aimed where all the buzzwords are in the market right now. The strategy, according to Ed Turkel, aligns the company with requirements from the HPC, Web 2.0, and cloud computing communities. The initial announcement for the group is broad, bringing together several elements of the organization for a services and product offering that gets the company off to a good start with this effort.

First up is the formation of the Scalable Computing & Infrastructure group from two groups within HP: the HPC team with which we are all familiar, and HP's Scalable Datacenter Infrastructure team. The SDI team has spent the last two years out of the public eye at HP, focused exclusively on a small group of large media and internet services companies, like Weta Digital (the team that did special effects for the Lord of the Rings series), Fox Interactive Media, and snapfish.com (HP's photo sharing site). This team has spent the last two years working on custom solutions for the large infrastructure demands of companies facing the very large scale out computing demands of internet users. It brings a practical focus on those needs to HP's new efforts, along with a select group of marquee customers with, one presumes, deep pockets.

On the hardware side, HP is announcing the Xeon-based HP ProLiant BL2x220c G5 blade. This blade allows HP to cram over 12 TFLOPS (1,024 quad-core sockets) into a single 42U rack, a very dense solution. This density comes with a lot of engineering and at the price of some functionality. Of course HP has put in their custom fan kit, and the BladeSystem that holds the new C-class compliant G5 blade is engineered for effective (air) cooling and does some smart power supply management to keep operations at the knee of the efficiency curve.

HP gets two dual-socket servers on a single blade by carefully selecting what makes it in the design and what gets left out. You can choose from the dual-core Xeon 5200 or quad-core Xeon 5400, each with 4 DDR2 DIMMs per blade. HP saves power using DDR2 instead of the FBDIMMs more commonly seen in Intel-based servers and by having 4 DIMM slots instead of 8. Each server also has only one PCI-Express mezzanine socket and one disk drive. These design tradeoffs obviously mean that the new blade isn't the right compute foundation for every task -- in particular a maxed out quad-core solution would be light on memory per core -- but HP is very specifically focusing this product for the customer that wants a lot of compute in a small space.

HP is also rolling its previously announced HP StorageWorks 9100 Extreme Data Storage System (ExDS9100) into this offering. Turkel described the ExDS9100 as "scale out" storage, offering lots of capacity without the expense and operational overhead of the reliability features that typical high-end, back office, never-go-down storage systems have to provide.

In a nod to HP's awareness of the needs of this market in which many customers are new to big compute, the company is turning to the facility expertise gained in its recent acquisition of EYP Mission Critical Facilities to form the foundation for its Data Center Transformation Services offering. This service is designed to help customers design and implement the compute infrastructure from top-to-bottom.

So where is the company headed long term? Is this new focus on cloud and internet-scale computing just the beginning of a change in focus for the company? "This is not the beginning of a strategy for the provision of cloud computing services from HP," said Turkel. "Rather, our focus is on helping customers build out the datacenter piece of cloud and other major computational infrastructure operations." In HPC, HP is ultimately looking to grab market from its competitors. According to Turkel, "Our goal is to add to our presence in the TOP500 list by capturing more of the scale out market."

HPCwire on Twitter

Discussion

There are 0 discussion items posted.

Join the Discussion

Join the Discussion

Become a Registered User Today!


Registered Users Log in join the Discussion

May 18, 2012

May 17, 2012

May 16, 2012

May 15, 2012

May 14, 2012

May 11, 2012

May 10, 2012

May 09, 2012

May 08, 2012


Most Read Features

Most Read Around the Web

Most Read This Just In

Arkeia

Around the Web

NVIDIA’s Bill Dally Talks 3D Chips and More at GTC

May 16, 2012 | Chief scientist discusses memory stacks, interconnects, and US technology leadership.
Read more...

NVIDIA Unveils Virtualized GPU with Kepler-Based Board

May 15, 2012 | GPU maker conjures up visualization technology for virtual desktops.
Read more...

Zettaflops Will Happen Says HPC Analyst

May 14, 2012 | Pessimistic predictions about technology have a poor track record, according to 451's John Barr.
Read more...

Next-Gen Memory on the Horizon

May 10, 2012 | DRAM manufacturers gear up for DDR4.
Read more...

US Energy Secretary Talks Supercomputing

May 09, 2012 | Steven Chu discusses the role of supercomputing in energy research.
Read more...

Sponsored Whitepapers

Sponsored Multimedia

ISC Think Tank 2012

Newsletters



HPC Job Bank


Featured Events







HPC Wire Events