Amazon Web Services has announced Cluster Compute Instances for Amazon EC2, a new instance type designed for high-performance computing (HPC) applications and other demanding network-bound applications.
In a July 13 press release on the new offering, Amazon said customers with complex computational workloads such as tightly coupled parallel processes, or with applications sensitive to network performance, can now achieve the same high compute and networking performance provided by custom-built infrastructure while benefiting from the elasticity, flexibility and cost advantages of Amazon EC2.
"Businesses and researchers have long been utilizing Amazon EC2 to run highly parallel workloads ranging from genomics sequence analysis and automotive design to financial modeling," said Peter De Santis, general manager of Amazon EC2, in a statement. "At the same time, these customers have told us that many of their largest, most complex workloads required additional network performance. Cluster Compute Instances provide network latency and bandwidth that previously could only be obtained with expensive, capital intensive, custom-built compute clusters. For perspective, in our last pre-production test run, we saw an 880 server sub-cluster achieve a network rate of 40.62 TFlops - we're excited that Amazon EC2 customers now have access to this type of HPC performance with the low per-hour pricing, elasticity, and functionality they have come to expect from Amazon EC2."
Indeed, Amazon officials said that before the company produced Cluster Compute Instances for Amazon EC2, enterprises with advanced HPC needs typically had to fund expensive, in-house compute clusters by purchasing dedicated, purpose-built hardware. However, with Cluster Compute Instances, users have access to the high-performance computing capabilities they need - with pay-as-you-go pricing, the ability to scale on-demand, and no upfront investments.
Amazon officials said Cluster Compute Instances provide more CPU than any other Amazon EC2 instance. Customers can also group Cluster Compute Instances into clusters to enable applications to get the low-latency network performance required for tightly coupled, node-to-node communication that is typified by many HPC applications. Moreover, depending on usage patterns, applications can see up to 10 times the network throughput of the largest current Amazon EC2 instance types.
"Many of our scientific research areas require high-throughput, low-latency, interconnected systems where applications can quickly communicate with each other, so we were happy to collaborate with Amazon Web Services to test drive our HPC applications on Cluster Compute Instances for Amazon EC2," said Keith Jackson, a computer scientist at the Lawrence Berkeley National Lab, which features the Department of Energy's National Energy Research Scientific Computing Center (NERSC). "In our series of comprehensive benchmark tests, we found our HPC applications ran 8.5 times faster on Cluster Compute Instances for Amazon EC2 than the previous EC2 instance types."
"The high-performance networking of Cluster Compute Instances for Amazon EC2 fills an important need among scientific computing professionals, making the on-demand and scalable cloud environment more viable for technical computing," said David Patterson, an expert on HPC and co-inventer of RAID, RISC and other computer innovations.
Cluster Compute Instances complement other AWS offerings designed to make large-scale computing easier and more cost effective, Amazon officials said. For example, Public Data Sets on AWS provide a repository of useful public data sets that can be easily accessed from Amazon EC2, allowing fast, cost-effective data analysis by researchers and businesses, Amazon said in its press release. These large data sets are hosted on AWS at no charge to the community. Additionally, the Amazon Elastic MapReduce service enables low-friction, cost effective implementation of the Hadoop framework on Amazon EC2. Hadoop is a popular tool for analyzing very large data sets in a highly parallel environment, and Amazon EC2 provides the scale-out environment to run Hadoop clusters of all sizes.
To get started using Cluster Compute Instances for Amazon EC2, visit http://aws.amazon.com. More information on Amazon EC2 and Cluster Compute Instances can be found at http://aws.amazon.com/hpc-applications.