April 27, 2012
Penguin Computing is flapping its wings, as it has received Intel's Data Center Innovation Award at the Intel Solutions Summit.
A press release describes the award as going to a company that exhibits "… successful deployment and integration of a data center or server solution resulting in superior return on investment for the client." This year, the Penguin on Demand (POD) cloud service gained that recognition from Intel.
We spoke with Arend Dittmer, director of product marketing at Penguin, about the news and the company's cloud offerings.
"If you do the math [POD] provides an enormous ROI for customers in different scenarios," he explained, "For example, customers that can't really afford to deploy an in-house HPC infrastructure – it makes so much more sense for them to use high performance computing on a pay-as-you-go basis."
The HPC on-demand model has seen some high profile use lately. Cycle Computing recently spun up a 50,000-core cluster over a three-hour time span, using Amazon's service. Computational chemists used the resources to identify suitable drug compounds for cancer research. Total cost of operation ran just shy of $5,000.
Penguin's cloud operations currently occupy datacenters in Salt Lake City, Mountain View and Indiana University. The latter, dubbed the academic cloud, is used strictly for government and academic research. Between its three facilities, the POD service has roughly 3,000 cores available for compute jobs. While this infrastructure is far smaller than Amazon's, Dittmer provided a few key differences between Penguin and its competitor.
"We actually run all HPC compute nodes on bare metal, without a virtualization layer involved. So customers don't have to allocate compute instances and tear them down when they're done," he said.
The service uses virtualization to provide access to gateway nodes, but all the jobs are bare metal.
Beyond providing on-demand compute cycles, Penguin has taken the concept further and offered on-demand application licenses as well. In this case, a cloud user would have a single point of payment for cycle and application usage. By making this option available, Penguin can offer applications that provision licenses to specific hardware.
When asked about use cases for the application-on-demand service, Dittmer mentioned IMMI, a company that manufactures safety systems. To model products like rollover cages for semi-trucks, IMMI uses POD to provide compute power as well as instances of LS-Dyna, an application used to simulate vehicle crashes.
One of the most interesting features about Penguin's cloud offering comes from their turnkey systems. The company makes the POD service available to users of in-house clusters in what was described as transparent offloading of excess workloads.
So far, Penguin on Demand has processed more than 15 million compute jobs for organizations including Dolby, Fluid Inc., IMMI and the California Institute of Technology. Increased focus on manufacturing and materials discovery may result in higher demand for HPC resources. Cloud offerings present an attractive option for organizations looking to avoid up front capital expenses.
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?