I have to say I'm fairly excited at the news today that Amazon is making available a new instance type for Amazon's Elastic Compute Cloud (EC2) service called Cluster Compute Instances. Specifically designed for high performance computing (HPC) applications, I can how this new service will be attractive to new start-ups who need HPC but don't want to spend the considerable money on their infrastructure just yet.
One particular area which I have experience in that uses HPC is the vertical search niche. Crawling and indexing large amounts of data is CPU and RAM intensive and even with servers and in particular storage prices continually dropping it is still costly to setup, maintain and deploy your own crawling cluster. For start-ups this will make Amazon's Cluster Compute Instances very attractive indeed. I will have to run some numbers on the costs of using Amazon's Cluster Compute Instances compared to an in-house setup to see how it stacks up. In the meantime here's the specs on what you get with each Cluster Compute Instance:
The Cluster Compute instance family currently contains a single instance type, the Cluster Compute Quadruple Extra Large with the following specifications:
23 GB of memory
33.5 EC2 Compute Units (2 x Intel Xeon X5570, quad-core "Nehalem" architecture)
1690 GB of instance storage
64-bit platform
I/O Performance: Very High (10 Gigabit Ethernet)
API name: cc1.4xlarge
Amazon allows you deploy up to 8 of their Cluster Compute Instances which are available in the US - N. Virginia Region on the Linux platform. If you need more than 8 Instances then you have to contact Amazon directly to request this.
I'm looking forward to giving Cluster Compute Instances a try at some point.
Leave a comment - Sign in with SpaceRef, Google, Yahoo or OpenID accounts