Digital Edition

SYS-CON.TV
Rackable Slams Traditional Blades as Wrong-Headed; Shows 'em How To Do It
Rackable Slams Traditional Blades as Wrong-Headed; Shows 'em How To Do It

Rackable Systems Inc, the little company getting big in large-scale data center deployments, claims that the blade merchants have been doing it all wrong all this time and that its latest development, a thing called the Scale Out Server, is the first architecture that finally gets it right - at least for large, high-end, high-density server farms.

First of all, the Scale Out Server, which will be trotted out at LinuxWorld, is based on bog-standard, investment-protecting, commodity parts - not proprietary lock-in designs like Rackable claims IBM's Blade Center used to avoid cannibalizing IBM's 1U business.

Scale Out lets the customers choose whatever motherboards, "real" processors - not notebook ones, God knows - networking, local storage, storage interconnects and provisioning, clustering and high-availability software they want.

The stuff will also run Linux, Windows, Solaris x86 and BSD though 80% of Rackable's customers run Linux, mostly Red Hat, with a rising amount of SuSE.

Rackable calls its approach "open blade."

Rackable also contends that the IBM Blade Center, which has got 14 blades squeezed in a 7U chassis, isn't suited to big deployments. It says this kind of design, which is typical of traditional blade servers, inhibits an open architecture and severely hampers thermal efficiency. It dismisses it as being for small, undemanding, risk-averse applications.

Rackable believes its new widget combines the industry's highest server density with the maximum computer power. It can stuff 92 one- or two-way blades or compute nodes - a total of 184 Xeons, say - in a rack that provides 10U of space for networking equipment or alternate network architectures.

The box, by the way, which features plug-and-play modularity and backplane mating, is supposed to be turnkey.

The company says it can build Scale Out Servers using P4s, Xeons, Noconas, Itaniums and Opterons.

Rackable's nodes are laid out side-by-side and back-to-back for thermal management and power efficiency. It says it's the most practical and cost-effective approach to high density and makes servicing the creature a lot easier.

The company figures that between having the I/O in front and the way it's cleaned up the cabling - 100% internalized cables - reduces maintenance time by up to 75%.

Rackable's got patented cooling technology that it says directs 92% of chassis' airflow to cooling the processors, which is what lets it put all those very hot chips - stuff that throws off 129 watts each, for Pete's sake - in such tight quarters without melting down.

Scale Out also comes with optional DC power - a telco trick - for maximum power savings, reducing the heat by maybe another 20%-40% and cutting power costs by maybe as much as 30%, the company figures.

Rackable fancies Scale Out Server for massive systems, basically your HPC segment, given to hard-core compute-intensive applications such as digital content creation and image rendering, biotechnology and pharmaceutical analysis, trading and portfolio analysis, web server farms, gas and oil exploration and scientific research labs.

There is, it says, growing demand among the Fortune 500s and enterprise crowd for such things running Oracle9i RAC and the new Oracle 10g.

Rackable says Scale Out Server, optimized for cabinet-level deployments, is especially well suited to applications of upwards of a thousand nodes.

Rackable has seed quantities of the thing out with early adopters. Pricing depends on configuration and the amount of RAM spec'd. A P4 node can run under $1,000. An average dual-processor node runs $2,800-$3,000. Fully fleshed out, one could be talking real money.

A typical node could have two 3.5-inch drives or four 2.5-inch ones, a full-height PCI-X slot, two gigE and one serial management port. Total lights-out remote management is provided.

About Maureen O'Gara
Maureen O'Gara the most read technology reporter for the past 20 years, is the Cloud Computing and Virtualization News Desk editor of SYS-CON Media. She is the publisher of famous "Billygrams" and the editor-in-chief of "Client/Server News" for more than a decade. One of the most respected technology reporters in the business, Maureen can be reached by email at maureen(at)sys-con.com or paperboy(at)g2news.com, and by phone at 516 759-7025. Twitter: @MaureenOGara

In order to post a comment you need to be registered and logged in.

Register | Sign-in

Reader Feedback: Page 1 of 1

Matt Hammel is one of the most devious untrustable so called IT solutions providers out there. He has caused more problems for many small medium companies in the Los Angeles area then myself and many other can count. Penguin fired him as did Rackables. If you start with this guy you'll be packing up your office and moving onto a new job.

Rackable claims to have the "industry's highest server with the maximum computer power. It can stuff 92 one or two-way blades or compute nodes - a total of 184 xeons, say - in a rack". In the same same rack space (Rackable uses a 60U rack for this config), Penguin can fit 156 servers or 312 xeons with 8U left over using the Bladerunner. Now that is dense!!!

Can anyone say B.S.? I just visited Rackable's website and found no information on a blade solution. It's even funnier that Rackable slams other vendors but is using the same message that companies like Verari Systems, Appro, and Linux Networx have been saying all along. There is no way that Rackable's solutions can manage a new Nocona processor. They run way to hot. They have good solutions, granted, but this article makes me laugh.




ADS BY GOOGLE
Subscribe to the World's Most Powerful Newsletters

ADS BY GOOGLE

Containers and Kubernetes allow for code portability across on-premise VMs, bare metal, or multiple ...
"Akvelon is a software development company and we also provide consultancy services to folks who are...
Enterprises are striving to become digital businesses for differentiated innovation and customer-cen...
DXWordEXPO New York 2018, colocated with CloudEXPO New York 2018 will be held November 11-13, 2018, ...
In this presentation, you will learn first hand what works and what doesn't while architecting and d...
Most DevOps journeys involve several phases of maturity. Research shows that the inflection point wh...
Modern software design has fundamentally changed how we manage applications, causing many to turn to...
In his session at 20th Cloud Expo, Scott Davis, CTO of Embotics, discussed how automation can provid...
In his session at 20th Cloud Expo, Mike Johnston, an infrastructure engineer at Supergiant.io, discu...
@DevOpsSummit at Cloud Expo, taking place November 12-13 in New York City, NY, is co-located with 22...
Founded in 2000, Chetu Inc. is a global provider of customized software development solutions and IT...
DXWorldEXPO | CloudEXPO are the world's most influential, independent events where Cloud Computing w...
DevOpsSummit New York 2018, colocated with CloudEXPO | DXWorldEXPO New York 2018 will be held Novemb...
CloudEXPO New York 2018, colocated with DXWorldEXPO New York 2018 will be held November 11-13, 2018,...
Dynatrace is an application performance management software company with products for the informatio...
Disruption, Innovation, Artificial Intelligence and Machine Learning, Leadership and Management hear...
SYS-CON Events announced today that DatacenterDynamics has been named “Media Sponsor” of SYS-CON's 1...
SYS-CON Events announced today that IoT Global Network has been named “Media Sponsor” of SYS-CON's @...
SYS-CON Events announced today that CrowdReviews.com has been named “Media Sponsor” of SYS-CON's 22n...
A valuable conference experience generates new contacts, sales leads, potential strategic partners a...