Intel & Low Latency
Read Intelligence in Finance, Intel’s new quarterly newsletter for financial services.
Intel defines a platform as a node, where a node is a stand alone computing resource. A Low Latency Platform is a connected set of resources that provides very fast response time.
In financial services, Latency is discussed in two ways:unidirectional latency and round trip latency.Capital markets firms closely guard their “Latency Intellectual Property” and speeds, but a typical roundtrip solution is about 30 milliseconds.
An example of unidirectional latency for capital markets is the time it takes for data to travel from a Wall Street Trader’s workstation to a Stock Exchange. Once the trader hits the “Execute Trade” button, the data leaves the trader’s workstation, passing through order management systems to the exchange. There are several functional areas that can consume significant amount of computing resources and introduce latency.
One bulge bracket firm has stated that lowering their latency by 10 milliseconds is worth a $200 million investment.
Low Latency does not perfectly overlap with Grid or High Performance computing. However, many complex algorithmic trading solutions, are certainly both Grid based and Low Latency.
This online resource, sponsored by Intel, includes news on technologies and vendors offering low latency solutions, case studies of rollouts of low latency architectures, and resources such as datasheets and benchmarks.
