NextIO: Leading Provider of Next Generation I/O Solutions

Financial Services

The compression of systems into dense rack-mounted and blade servers only confirms the need for ever increasing compute power in the data center. This need is even greater for numerical analysis needed in financial modeling applications. Though density keeps increasing in standard x86 servers, the needs of these applications are so great that dedicated, specialized servers are often used.

Dedicated systems such high-powered, multi-core, multi-socket processor servers or blades required for financial analysis and financial modeling calculations expel a tremendous amount of heat and require hundreds of watts of power. They tend to be more expensive, and potentially require additional installation, configuration, and maintenance. Since these are dedicated systems, they cannot be utilized and amortized across a broad set applications to offset the cost. The high cost of data center real-estate along with the costs associated with maintaining and operating more specialized computing platforms makes having machines dedicated to specialized applications such as financial analysis an expensive and time-consuming proposition.

Substantial Performance Increase at a Fraction of the Cost

The issues surrounding large scale financial analysis can be addressed using standard General Purpose Computing using Graphics Processing Units (GPGPU) cards readily available today. By taking advantage of the inherent numerical processing requirements needed for graphics and offloading the binomial calculations from the standard server architecture processor to the GPGPU you can increase the performance of the calculations per second substantially. The NextIO Adaptive Connect™ platform provides the ideal solution for applications that require standard off-the-shelf PCI Express I/O devices that cannot be easily integrated in servers such as GPGPUs (see Figure 1).

NextIO's Adaptive Connect platform delivers 100X performance over standard x86 server I/O models at a fraction of the cost. Any dense computing environment that requires GPGPUs for financial or numerical analysis or vector processing can now use off-the-shelf hardware. The initial capital expenditure is much lower, and since these are generic, commodity systems the operational costs including maintenance, power, cooling, and space are also significantly lower. With standard machines, data centers enjoy the added benefits of repurposing these machines for other tasks when not in use, increasing the overall utilization of the infrastructure.

Seamless Scalability

The ability to add new specialized hardware into an existing data center without disrupting the existing infrastructure is valuable in its own right. The ability to integrate off-the-shelf components to address a specialized problem without requiring major changes to a data center and to implement upgrades as advances are made in the GPGPU arena without restructuring is of significant importance. The NextIO solution, leveraging PCI Express, offers today's financial services organizations an opportunity to rapidly increase their capacity without additional costly investments. The need for expensive dedicated servers to handle tasks that can be offloaded to the GPGPU's ensures that the investments made into the existing data center are not lost; in fact they are given renewed life with the ability to easily add upgrades and additional I/O capacity by utilizing the NextIO product family.