[Beowulf] $1, 279-per-hour, 30, 000-core cluster built on Amazon EC2 cloud
prentice at ias.edu
Mon Oct 3 10:51:06 PDT 2011
Thanks for posting that video. It confirmed what I always suspected
about clouds for HPC.
On 10/03/2011 08:25 AM, Douglas Eadline wrote:
> Interesting and pragmatic HPC cloud presentation, worth watching
> (25 minutes)
>> $1,279-per-hour, 30,000-core cluster built on Amazon EC2 cloud
>> By Jon Brodkin | Published September 20, 2011 10:49 AM
>> Amazon EC2 and other cloud services are expanding the market for
>> high-performance computing. Without access to a national lab or a
>> supercomputer in your own data center, cloud computing lets businesses
>> up temporary clusters at will and stop paying for them as soon as the
>> computing needs are met.
>> A vendor called Cycle Computing is on a mission to demonstrate the
>> of Amazonâs cloud by building increasingly large clusters on the Elastic
>> Compute Cloud. Even with Amazon, building a cluster takes some work, but
>> Cycle combines several technologies to ease the process and recently used
>> them to create a 30,000-core cluster running CentOS Linux.
>> The cluster, announced publicly this week, was created for an unnamed
>> âTop 5
>> Pharmaâ customer, and ran for about seven hours at the end of July at a
>> cost of $1,279 per hour, including the fees to Amazon and Cycle Computing.
>> The details are impressive: 3,809 compute instances, each with eight cores
>> and 7GB of RAM, for a total of 30,472 cores, 26.7TB of RAM and 2PB
>> (petabytes) of disk space. Security was ensured with HTTPS, SSH and
>> AES encryption, and the cluster ran across data centers in three Amazon
>> regions in the United States and Europe. The cluster was dubbed
>> Spreading the cluster across multiple continents was done partly for
>> recovery purposes, and also to guarantee that 30,000 cores could be
>> provisioned. âWe thought it would improve our probability of success if
>> spread it out,â Cycle Computingâs Dave Powers, manager of product
>> engineering, told Ars. âNobody really knows how many instances you can
>> get at
>> any one time from any one [Amazon] region.â
>> Amazon offers its own special cluster compute instances, at a higher cost
>> than regular-sized virtual machines. These cluster instances provide 10
>> Gigabit Ethernet networking along with greater CPU and memory, but they
>> werenât necessary to build the Cycle Computing cluster.
>> The pharmaceutical companyâs job, related to molecular modeling, was
>> âembarrassingly parallelâ so a fast interconnect wasnât crucial. To
>> reduce costs, Cycle took advantage of Amazonâs low-price âspot
>> instances.â To
>> manage the cluster, Cycle Computing used its own management software as
>> as the Condor High-Throughput Computing software and Chef, an open source
>> systems integration framework.
>> Cycle demonstrated the power of the Amazon cloud earlier this year with a
>> 10,000-core cluster built for a smaller pharma firm called Genentech. Now,
>> 10,000 cores is a relatively easy task, says Powers. âWe think weâve
>> the small-scale environments,â he said. 30,000 cores isnât the end
>> either. Going forward, Cycle plans bigger, more complicated clusters,
>> ones that will require Amazonâs special cluster compute instances.
>> The 30,000-core cluster may or may not be the biggest one run on EC2.
>> isnât saying.
>> âI canât share specific customer details, but can tell you that we do
>> businesses of all sizes running large-scale, high-performance computing
>> workloads on AWS [Amazon Web Services], including distributed clusters
>> the Cycle Computing 30,000 core cluster to tightly-coupled clusters often
>> used for science and engineering applications such as computational fluid
>> dynamics and molecular dynamics simulation,â an Amazon spokesperson told
>> Amazon itself actually built a supercomputer on its own cloud that made it
>> onto the list of the worldâs Top 500 supercomputers. With 7,000 cores,
>> Amazon cluster ranked number 232 in the world last November with speeds of
>> 41.82 teraflops, falling to number 451 in June of this year. So far, Cycle
>> Computing hasnât run the Linpack benchmark to determine the speed of its
>> clusters relative to Top 500 sites.
>> But Cycleâs work is impressive no matter how you measure it. The job
>> performed for the unnamed pharma company âwould take well over a week
>> them to run internally,â Powers says. In the end, the cluster performed
>> equivalent of 10.9 âcompute years of work.â
>> The task of managing such large cloud-based clusters forced Cycle to step
>> its own game, with a new plug-in for Chef the company calls Grill.
>> âThere is no way that any mere human could keep track of all of the
>> parts on a cluster of this scale,â Cycle wrote in a blog post. âAt
>> weâve always been fans of extreme IT automation, but we needed to take
>> to the next level in order to monitor and manage every instance, volume,
>> daemon, job, and so on in order for Nekomata to be an efficient 30,000
>> tool instead of a big shiny on-demand paperweight.â
>> But problems did arise during the 30,000-core run.
>> âYou can be sure that when you run at massive scale, you are bound to
>> into some unexpected gotchas,â Cycle notes. âIn our case, one of the
>> included such things as running out of file descriptors on the license
>> server. In hindsight, we should have anticipated this would be an issue,
>> we didnât find that in our prelaunch testing, because we didnât test
>> at full
>> scale. We were able to quickly recover from this bump and keep moving
>> with the workload with minimal impact. The license server was able to keep
>> very nicely with this workload once we increased the number of file
>> Cycle also hit a speed bump related to volume and byte limits on
>> Elastic Block Store volumes. But the company is already planning bigger
>> better things.
>> âWe already have our next use-case identified and will be turning up the
>> scale a bit more with the next run,â the company says. But ultimately,
>> not about core counts or terabytes of RAM or petabytes of data. Rather,
>> about how we are helping to transform how science is done.â
>> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
>> To change your subscription (digest mode or unsubscribe) visit
>> This message has been scanned for viruses and
>> dangerous content by MailScanner, and is
>> believed to be clean.
More information about the Beowulf