So the dictionary definition of latency is fairly straightforward, but its generality highlights some of the challenges in defining latency within trading systems: what stimuli and responses are we talking about? The reality is that any trading system is composed of multiple components, and each of those components receives stimuli (input data) and produces responses (output data). I've written in the past about the need to understand the holistic latency through all of these systems, but I thought it would be useful to identify the different categories of latency that tend to exist, in order that people can start to think about the potential solutions that are available for each of them.
Categorisations are always difficult and potentially subjective (see Linnaeus as a good example of the challenges), but in my mind there are four distinct areas that need to be considered when thinking about latency in trading applications, and each of them has a number of different ways of contributing to holistic latency. Sumerian's CTO Peter Duffy takes an indepth look at the various different types of latency that need to be considered when thinking about latency in trading applications.
Read the full article on the Sumerian blog>