Electricity supplier Haven Power is using the Amazon Web Services (AWS) cloud to provide cost-effective access to more computing power than it will ever need as the firm grows rapidly.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
The company originally moved to a cloud-based model for its disaster recovery but soon began reaping other benefits of a service offering on-demand computing resources that are charged on a pay-as-you-go basis.
Haven Power, which supplies businesses with electricity. has grown from its four founders to 350 people since 2006. However, IT has not kept pace.
Paul Armstrong, business systems manager at Haven Power, said the technology at the firm had not kept up with its growth: "I was brought in to look at everything in the IT estate."
The company's IT, which had no virtualisation, was made up of a mixture of onsite and offsite servers, with a billing platform hosted in the datacentre of its parent Drax Group.
Armstrong said the company had no disaster recovery at the time, which became the overriding concern that had to be addressed quickly.
Haven was also embarking on a major programme to change billing systems and needed access to more computing resources for development and testing. "We were running out of disc space at a time when we were running a major change programme," said Armstrong.
He says the company's first priority was to get disaster recovery in place and that it had three choices about how to do this.
The firm sent a request for proposals (RFP) to three organisations. The first option was to get disaster recovery though the datacentre of its parent company Drax Group, which operates a power station. The second option was for a re-located disaster recovery service through a third-party, ICM, and the final option was a cloud service.
The cloud option RFP was sent to IT consultancy Smart421, which was subsequently chosen to help select and integrate the cloud. During the planning process Microsoft's Azure cloud computing technology was considered but Haven selected Amazon.
"We looked at Azure and although it is very powerful we did not think it was mature enough and we were concerned about lock-in," said Armstrong.
The tenders were sent out in January and February this year. In parallel and into April, Haven worked on its IT infrastructure to make it more robust. This included the introduction of server virtualisation. Smart421 then designed the overall solution to enable the use of AWS.
The company first put its disaster recovery into the AWS cloud but is now harnessing it for testing and development. Haven can access the Amazon cloud when it needs compute power for short periods and pay for only what it uses. "We are in a position now where we have an on-premise capability as well as a cloud infrastructure, through AWS," said Armstrong.
The power of the cloud was recently demonstrated to Haven Power when it had an issue with a pricing server.
"When we had a problem with a system, I was telephoned, while travelling from Ipswich to London, by IT staff to ask whether they should build a replacement. We agreed to do this using the AWS cloud and by the time I arrived in London (about one-and-a-half hours) it was complete."
He said it would normally take a few days to complete such a task if the company had the computing capacity available in-house, and if it didn't the process would take a week.
AWS's cloud hosts a total of 566 billion objects ranging from Word documents to huge video files. Amazon says it has lowered AWS prices 15 times in the last five years as the number of customers increases, offering flexible payment models to customers.
|Achieving rapid scale through the cloud with AWS|
|A pharmaceutical company was testing a new product and required the equivalent of a 30,000-node supercomputer to run molecular modelling. To be able to build a super computer it contacted a company known as Cycle Computing. The firm had to launch the system across three Amazon Web Services regions simultaneously - US West (California), US East (Virginia) and EU West (Ireland). There was 30TB of memory available on the compute nodes, which could hold the entire raw Wikipedia database five times over. To build a system of this scale traditionally would take many months and millions of dollars. By using AWS, Cycle Computing was able to build the supercomputer in hours and at a cost of less than $10,000, or $1,279 per hour.|