Specifications

4
High Performance Trading/Algo Speed with Wombat Design and Implementation Guide
OL-15617-01
Introduction
Target Market
Financial services firms require vast amounts of computing power to run their business. They use
homegrown applications or increasingly ISV applications to do these computations. However, these
applications were originally built for SMP machines or hard-coded clusters of a certain size. The net
result is that computations takes hours and days, while the business needs it to happen in seconds and
minutes.
Further, the average utilization of the server farm is very low ( less than 10 percent). Given a fixed IT
budget, this means the firm does a lot less computation than it would like to. Fewer scenarios are run and
modeling is done to less accuracy. Some businesses do not use modeling at all due to lack of resources
Solving these problems can provide a competitive advantage to the customer. Specifically, the following
businesses are the ones most affected:
• Front office—Pricing and hedging of derivatives, foreign exchange options, and other structured
financial products
• Mid office—Risk analysis of portfolio, counter-party credit risk, enterprise risk, proprietary desk
analytics
• Back office—Fraud detection, global treasury
Automated Trading Benefits
Automated trading benefits include the following:
• Delivers low-latency and high speed Ethernet and InfiniBand interconnects for financial
applications such as trading floors and market data feed
Automated trading makes reducing latency and increasing performance essential in a market data
environment. Feed handlers receive real-time market data feeds from sources such as Options Price
Reporting Authority (OPRA), NASDAQ, and electronic communications networks (ECNs), and
these feeds need to be “normalized” before being distributed to users. Data normalization allows
feeds to be standardized through dedicated feed processors and entered into a uniform database
model. The uniform access to multiple normalized market data feeds facilitates data distribution to
end-users and ensures data consistency throughout the organization. The high server-to-server
traffic that occurs as these computations are performed means that a cluster of servers with the
lowest possible latency interconnect is needed to reduce delay in delivering the market data.
Through the use of RDMA technology, an application can offload all communications management
to the InfiniBand host channel adapter, which allows more CPU cycles to be spent on processing,
rather than communications. Cisco’s innovative HPT solution creates a high-performance server I/O
fabric, achieving ultra-low-latency performance to support the growing computing needs of market
data feed handlers and other trading floor applications.
• Minimizes latency in each component of delivery platform
It is imperative that latency be minimized when delivering time-sensitive data. As the data traverses
the different components of a trading platform—including market data delivery, order routing, and
execution—an HPC environment addresses the speed-sensitivity requirements by providing a lowest
latency interconnect, so that raw computational power can be used in clusters to deliver the fastest
response possible.
• Helps prepare for new regulations that drives high market-data volumes