Entries filed under “Datacenter operations”

New items related to configuration of datacenters or the equipment in them, innovation in the deployment of power and cooling infrastructure, monitoring, and the operation of large scale datacenters

Video: Underground Datacenter in Helsinki

By Rich Brueckner | Published February 23, 2011

In this video, CNN looks at an underground datacenter built by Helsingen Energia and ISP Academica in Helsinki. Located 30 meters below a historic church, the 2 megawatt facility is cooled by sea water and its waste heat is used to heat residential buildings.

Eco-efficient computer halls could really save significant amounts of primary energy and at the same time recover wasted heat for recycling,” said Sipila. “It is perfectly feasible that a quite considerable proportion of the heating in the capital city could be produced from thermal energy generated by computer halls. When it’s cold outside, you need a lot of heat to warm the house. But the data centers were putting all of this heat into the atmosphere.”

At tip of the hat goes to Data Center Knowledge for pointing us to this story.

Also posted in HPC, Video | 2 Comments

Adaptive’s Moab Suites Adds Automated Provisioning

By Rich Brueckner | Published February 17, 2011

Automation in the datacenter is a bit of holy grail. The goal is to relieve system administrators from doing repetitive tasks so that they can focus on the hard stuff that helps their enterprise make money.

To this end, this week Adaptive Computing announced version 6.0 of their Moab Adaptive Compututing Suite this week. Charles Babcock over at InformationWeek writes that the upgrade enables policy-driven automated provisioning of virtualized resources in an enterprise data center.

Moab can recognize high-priority versus low-priority workload placement,” said Peter ffloulkes, Adaptive’s VP of marketing. “It can decide on the optimal placement of a new service on a server cluster, depending on the resources available.”

Also posted in HPC, HPC Software | Leave a comment

Duke Study Looks at Power Consumption Trends in HPC Clusters

By Rich Brueckner | Published February 14, 2011

John B. Pormann from Duke University has posted a two-year study of power consumption across 500-600 machines of the Duke Shared Cluster Resource (DSCR). The DSCR is a large Linux cluster – over 4100 CPU-cores – that provides “production-grade” computational capabilities for over 70 research groups and 700 users from all across campus.

As power and cooling become increasingly important to HPC installations, this data suggests that vendors such as Dell and Intel are responding in concrete and meaningful ways, with newer platforms that require less power while providing higher levels of performance.

Also posted in HPC | 2 Comments

The Days of Raised Floors are Numbered

By Rich Brueckner | Published February 4, 2011

Stephen Bigelow writes that raised floors may be going the way of the dodo.

According to TechTarget’s 2010 Data Center Decisions survey, 59% of IT respondents use raised flooring in their current data center, but only 43% expect to use raised floors in a future data center. Slabbed floors are also falling into disuse, with 33% of respondents using slabbed floors in the current data center, but only 19% planning slabbed floors for a future data center. In fact 38% of IT professionals don’t know what kind of flooring they will use in the future.

Bigelow goes on to say that while raised floors have been around for a long time, they just don’t provide enough cooling for today’s high-density, 42U rack enclosures. And having written my share of Cray site planning manuals, I can tell you that a dirty, cluttered underfloor full of cables and pipes makes a pretty lousy plenum for air to move through. Full Story

Also posted in HPC | 3 Comments

Datacenter Knowledge Survey Looks at 2011 Trends

By Rich Brueckner | Published January 21, 2011

What developments can we look forward to this year in the datacenter? Rich Miller at Datacenter Knowledge did a survey that revealed some surprising trends including low-power servers drawing as little as 30 Watts, pay as-you-use hardware acquisition models, and rapid adoption of modular datacenter units.

The scale and pace of required to meet data center demand is large and growing,” said Anthony Wanger, President of i/o Data Centers. “It is too large and fast-paced to be solved with series of ‘snowflakes’ – a model where every data center is uniquely designed, custom built and very expensive. For this reason, the era of the ‘data center as big construction project’ is over and is being replaced by the era of mass production.”

It all adds up to reducing the cost of a datacenter from today’s average of $10 Million per Megawatt to something more like $3 Million. That will require mass production and economies of scale. Who’d have thought that the next step of the Information Age is to learn the lessons of the Industrial Revolution?

Also posted in Cloud HPC, HPC | 1 Comment

Webcast: Oracle Grid Engine Product Road Map

By Rich Brueckner | Published January 14, 2011

In this webcast, Oracle’s Dan Templeton presents the key capabilities and product roadmap for Oracle Grid Engine, the popular resource brokering software previously available as open source from Sun Microsystems.

Dan was careful to point out Oracle’s Grid Engine Product Forum for the Grid Engine Community and their Incubator Project, which is designed to facilitate continued community code contributions.

Registration is required to view, but you can download the MP3 audio.

Also posted in HPC, HPC Software | 1 Comment

HPC Power and Cooling: Amps

By Rich Brueckner | Published January 7, 2011

Dell’s John Fragalla is out with Part III of his post on HPC Power and Cooling, this time talking about how to best provision for cluster amperage. The overall focus of this piece is that HPC system design is as much about determining the overall power and cooling aspects as it is about processors, interconnect, and IO.

In some cases, depending on the amps per rack, one might need multiple rack PDUs per rack to balance the load across circuit breakers and phases to ensure, under peak load when running Linpack, no breakers are exceeding the maximum specification.

In Part 4, John will cover Cubic Feet per Minute (CFM), which is a measurement of air flow a HPC system produces from component fans.

Also posted in HPC, HPC Hardware | Leave a comment

When Will it be Time to Upgrade to 40 GbE?

By RichB | Published December 14, 2010

David Gross writes that, with all the excitement over the new 40 GbE standard, the reality is that line rate is not as important a measure of network capacity as it used to be, especially with multi-gigabit port prices having more to do with transceiver reach, cabling options, and port densities than framing protocol.

Within the data center, which line rate you choose is increasingly falling back in importance to which transceiver you choose, what type of cabling, whether to aggregate at end-of-row or top-of-rack, and so forth. Moreover, as I wrote a few weeks ago, the most power efficient data center network is often not the most capital efficient. So rather than considering when 40 Gigabit Ethernet upgrades will occur, I think it’s more important to monitor what’s happening with average link lengths, the ratio of installed singlemode/multimode/copper ports, cross-connect densities in public data centers, the rate of transition to IPv6 peering, which can require power-hungry TCAMs within core routers, and especially whether price ratios among 850, 1310, and 1550 nanometer transceivers are growing or shrinking. So rather than wondering when 40G will achieve a 3x price ratio to 10G, it’s equally important to consider whether 10G, 1550nm transceivers will ever fall below 10x the price of 10G, 850nm transceivers.

Also posted in HPC, HPC Hardware | Leave a comment

Gartner Report: Data Growth is Biggest Datacenter Challenge

By Rich Brueckner | Published November 2, 2010

According to a new Gartner survey, data growth is the biggest data center hardware infrastructure challenge for large enterprises. In fact, more than half of respondents said they plan to expand capacity at existing datacenter sites by end of 2011.

“While all the top data center hardware infrastructure challenges impact cost to some degree, data growth is particularly associated with increased costs relative to hardware, software, associated maintenance, administration and services,” said April Adams, research director at Gartner. “Given that cost containment remains a key focus for most organizations, positioning technologies to show that they are tightly linked to cost containment, in addition to their other benefits, is a promising approach.”

As reported here by John West, Big Data is a huge issue for the HPC community as well.

Also posted in HPC, Storage | Leave a comment

Webinar: Eliminating Complexity in Managing GPU Clusters

By Rich Brueckner | Published October 29, 2010

This week Bright Computing announced that James River Technical will sell and support Bright Cluster Manager. James River focuses on the Education space, and this is looks like a good move for Bright Computing as the company seeks to gain traction for its Linux cluster management software.

As part of this new collaboration, James River will host a Nov. 10 webinar on Eliminating Complexity in Managing GPU Clusters. Topics include:

  • Key challenges in managing GPU clusters
  • Two models of GPU management: Tool-kit vs. Single Solution
  • How to configure a GPU system from “bare-metal” to functioning, high-performance cluster in less than one hour
  • How to scale to thousands of nodes, including systems comprising a mixture of GPUs and CPUs
  • How to seamlessly switch between CUDA versions

For more information on how to register, check out the James River webinar page.

Also posted in Cloud HPC, GPUs, HPC, System Management | 1 Comment

Rocky Mountain Super Center Welcomes New Board Member

By John Leidel | Published September 20, 2010

The Rocky Mountain Supercomputing Center announced a new member to their Board of Directors.  Susan L. Baldwin, Executive Director of Compute Canada, was named to the board.

Susan is an excellent addition to the RMSC Board because she shares our vision of bringing High Performance Computing technology to small- and medium-sized businesses,” said Earl J. Dodd, RMSC Executive Director.

RMSC and the State of Montana have seized the leadership position in harnessing HPC technology as an economic driver that makes businesses more competitive in the global market place,” said Susan Baldwin.

Under Baldwn’s direction, Compute Canada has integrated HPC resources from seven partner consortia across Canada to create a powerful and dynamic computational resource.  Compute Canada and the university-based regional HPC consortia provide for overall architecture and planning, software integration, operations and management, and coordination of user support for the national HPC platform.

Congrats to RMSC and Susan Baldwin.  For more info, read their full release here.

Also posted in Collaborations, Enterprise HPC | Leave a comment

University of Arkansas Receives $1.7 million for HPC Improvements

By John Leidel | Published August 26, 2010

The University of Arkansas at Fayetteville announced that it has received a $1.7million grant from the National Science Foundation to improve the facility that houses the supercomputers for the college.  The funds will be used to purchase air conditioners for the supercomputers and equipment that ensures that electricity can run around the clock, since many research projects require programs to run for several days and cannot sustain a power outage.

This National Science Foundation grant will allow Arkansas to move forward substantially in the area of research computing. It will provide the infrastructure that we need to house large-scale computers and storage that support several areas of science, and will improve our ability to compete nationally. We are delighted that the National Science Foundation has chosen to support this project,” said Amy Apon, Ph.D., University of Arkansas computer science professor and Director of the Arkansas High Performance Computing Center.

The funding will also help enhance the network capabilities available to UA researchers from the state’s high speed optical network.  The high speed optical network, the Arkansas Research and Education Optical network, allows researchers using computers for advanced scientific research at Arkansas’s four-year universities to have better access to the state’s supercomputing resources.  For more info, read their full release here.

Also posted in HPC Hardware, Network | Leave a comment

Purdue Throttles Power Based on Heat Loads

By John Leidel | Published August 23, 2010

According to an article in CampusTechnology online, Purdue University IT staff have developed methodologies in order to control the operational performance of machines based on the current data center heat load.  Over the course of a sweltering summer, the Rosen Center for Advanced Computing at Purdue experienced several planned and unplanned outages.  As a result, the IT staff had to develop a method to allow users access throughout the chiller downtime.

Purdue LogoPower outages are actually infrequent at the data center, [Patrick] Finnegan said. But he added that this summer, “due to some planned cooling system maintenance, coupled with the unusually hot summer, we have had some brief cooling outages.”

In both instances, the cause was a temporary capacity reduction in the campus chilled water supply.”

Temperature sensors in the data center kick off scripts that throttle the relative power utilization of the machines.  The article doesn’t go into technical specifics, but my geek radar says they’re likely tickling the CPU stepping via ACPI.

The program worked, and the datacenter didn’t overheat, so the process was a success. We actually were a bit surprised it worked so seamlessly,” said [Mike] Shuey. “It’s much better to have jobs run slowly for an hour than to throw away everyone’s work in progress and mobilize staff to try to fix things.”

For more info, check out their full article here.

Leave a comment

Microsoft parallel runtime expected to go commercial next year

By John West | Published August 23, 2010

ZDNet wrote last week about a new HPC offering that Microsoft is evidently planning on moving from research to commercial product in the coming months. The platform is called “Dryad”

Microsoft logoDryad is an ongoing Microsoft Research project dedicated to developing ways to write parallel and distributed programs that can scale from small clusters to large datacenters. There’s a DryadLINQ compiler and runtime that is related to the project. Microsoft released builds of Dryad and DryadLINQ code to academics for noncommercial use in the summer 2009.

Dryad stackAs you can see from the diagram (click for a larger view), there is a lot of technology in the platform, including a compiler, runtime, a new file system (TidyFS), and a scheduler (Quincy). “Nectar” is a set of data management tools

“In a Nectar-managed data center, all access to a derived dataset is mediated by Nectar. At the lowest level of the system, a derived dataset is referenced by the LINQ program fragment or expression that produced it. Programmers refer to derived datasets with simple pathnames that contain a simple indirection (much like a UNIX symbolic link) to the actual LINQ programs that produce them.”

According to ZDNet, Dryad was outed in a presentation this month with plans to offer a Community Technology Preview in November 2010 (announced with SC10, I’m guessing), with a final release for Windows HPC Server by next year.

Also posted in Computing Research, HPC Software, System Management, Tools | 1 Comment

Sun GridEngine, now 100% less free

By John West | Published August 20, 2010

Oracle logoOracle continues its drive to do away with Sun’s strategy of making money by adding value on top of open source tools provided free to the community. In fairness to Oracle it didn’t work all that well as a business model for Sun, which hemorrhaged money and never turned the strategy into any significant share of the HPC market. The latest victim is Sun’s popular job scheduling system, GridEngine.

This change actually happened back in June as far as we can tell, but it’s just popped up as a topic on various discussion boards. Randall Hand at Vizworld wrote up a nice summary of the change, which moves the free Sun GridEngine to the for-pay Oracle GridEngine plus a 90-day evaluation trial.

Oracle has “absorbed” Sun GridEngine internally and renamed it “Oracle GridEngine” (OGE) and placed it under a new license that restricts it to only 90-days of free usage in a “trial” arrangement.  From the 6.2U6 EULA:

As selected in your Entitlement, one or more of the following Permitted Uses will apply to your use of Software. Unless you have an Entitlement that expressly permits it, you may not use Software for any of the other Permitted Uses. If you don’t have an Entitlement, or if your Entitlement doesn’t cover additional software delivered to you, then such software is for your Evaluation Use.

(a) Evaluation Use. You may evaluate Software internally for a period of 90 days from your first use.

So HPC centers large and small using Sun GridEngine are going to have to start ponying up, or move to something else. Or both.

Also posted in Business of HPC, System Management | 5 Comments

Advertisement

Penguin Ad

View All Videos

insideHPC.com is a production of insideHPC, LLC. © 2006-2011 Sitemap