Amanda Dunn, director of business development, EMEA at PEER 1 Hosting tells DCM how Microsoft HPC and PEER 1 Hosting have brought hosted Supercomputing to the financial services sector The financial sector is a time-critical, data rich industry. The speed of data processing can literally cost millions in the fast-moving world of financial risk. There is no room for errors, which means this industry needs computing services that perform at high speeds with little or no down time. With technology evolving at an increasing rate, hosting companies and vendors including ISVs are now able to provide Supercomputing services that are capable of conducting lightning fast calculations on a flexible basis, eliminating the need to purchase expensive hardware and software.

PEER 1 Hosting has teamed up with Microsoft’s HPC (High Performance Computing) team to offer GPU cloud services that assist the financial sector. The Supercomputing service enables parallel computation for processing large volumes of quantitative data very quickly, which is crucial for any financial service business. Previously, powerful computers and expensive software would need to be purchased to obtain such a service, which eliminated competition from the majority of smaller companies as the price was simply too costly. Now that that this type of service is available in the cloud on a monthly pay-as-you-go service, it offers affordability and accessibility to a wider range of companies.

The service that is offered by Microsoft HPC and PEER 1 Hosting significantly reduces the processing time for customers in a number of verticals including digital content creation, engineering, research and finance computation. Customers can either use clustered ISV applications; their own in house parallel or commercial off-the-shelf application such as Microsoft Excel.  By combining its Supercomputing service with Microsoft’s Windows Server HPC 2008 R2, PEER 1 is able to support for example a scale out of Excel into its GPU cloud. This supercomputing service provides an excellent framework for management of data, creating policy reports and the construction of complex calculations.

For those opting to use Excel, the service uses Windows Server HPC 2008, enabling multiple instances of Excel to run on a Windows HPC cluster. It then combines with PEER 1 Hosting’s cloud service built entirely on NVIDIA Telsa GPU’s to deliver supercomputing performance in the cloud without having to provision multiple servers on site. The NVIDIA Tesla computing cards are embedded into servers at PEER 1 Hosting’s data centres in Fleet and Portsmouth, both connected by its wholly owned 10GB fibre network to LINX. The hosted GPU cluster enables dramatically quicker financial insight in portfolio modeling through the transition to energy efficient parallel computing power by bringing the performance of a small cluster to a workstation. With hundreds of processor cores and a standard C compiler that simplifies application development, Tesla cards scale to solve the world’s most important computing challenges more quickly and accurately.  When all of these aspects are combined, the hosted solution offers burst Supercomputing capabilities that dramatically reduce processing time on a cost effective, flexible basis.

Data security of cloud services is a key concern in the financial and corporate sector, but by provisioning the GPU cluster on dedicated systems within a private cloud, PEER 1 Hosting is able to address this. Carrier grade connectivity and firewalls inside the PEER 1 data centre POPs also offers a level of security far beyond the public cloud, matching the security infrastructure financial institutions will be running at their own sites.

As financial portfolios continue to grow and become more complex, they will require more intricate computing solutions that complete actions at rapid speeds. The calculation of models that contain thousands of individual assets can take minutes or even hours on a regular computer and can have a direct impact on an organisation’s performance. The financial sector is reliant on real-time data and needs to act spontaneously on reactive market conditions; the requirement for affordable, utility-based, high-performance computing has never been greater.