Network latency is a serious matter for many modern corporate applications – definitely a case of less is more
There was a time when network latency was a matter for techno-obsessives only, scarcely a real business issue at all. Now, the length of time it takes for data to get from point A to point B is a mission-critical matter for many organisations, with ramifications for everyone from the boardroom to the stock room.
When discussing the merits of a data network, bandwidth is usually taken as the best metric of performance. Bandwidth is a measurement of the capacity of a network connection, typically expressed in terms of bits per second. Although important, bandwidth is only one measure that you need to think about when weighing up whether a network has the credentials for the job in hand.
Get me to the server on time
Latency is a measure of the time delay in processing network traffic, being the total time it takes for a network packet to travel from one server, through a network adapter, over a fibre connection, through an adapter at the other end and finally onto another server. Latency is determined not only by the thickness of the pipe in the middle, but also by the distance between points on the network, the choice of operating system, and the quality of devices through which the data must pass.
Naturally, high latency is bad (very bad if you’re talking about voice traffic or time-sensitive applications), while low latency is good. Excessively high latency can cause bottlenecks of data that prevent the main pipe from being used to its full potential, effectively lowering bandwidth. High latency can manifest for just a few seconds here or there, when it’s described as ‘temporary latency’, or is pretty much always there, which is ‘constant latency’.
Variations in latency or ‘jitter’ can have their own negative effect on the speed of a data packet or the quality of a phone conversation. You can solve some of these problems by increasing bandwidth, but all you’re really doing is staving off the inevitable need to find a lower latency alternative.
Every millisecond counts
An example of the network latency issue at its most critical is in the financial services sector, where trading firms are always looking for a competitive advantage based on tiny fractions of the speed of a transaction.
Everything changed a couple of years ago when the UK’s financial services regulator altered the way stock transactions work with a directive called MiFID. This meant trades no longer had to go through the London Stock Exchange, but could use any one of a number of competitive exchanges around the country, which are all now vying for business. If you want to trade with all of these exchanges, then you need a network solution that offers mega-low latency so you can engage in arbitrage between them.
A modern arbitrage strategy means you’re dealing with market opportunities that may only be available for a few milliseconds before an advantage is lost. With the appropriate algorithmic trading tool – and the right network – you can capitalize on minute differences in network speed and make a lot of money – or lose it if you’re on the wrong end of a better network.
Media companies, like broadcasters and advertising agencies, also rely on the fast transfer of large volumes of rich data between sites, usually to insanely tight deadlines. Retailers and manufacturers have supply chains where seconds count. Latency, allied to security, availability and reliability, is an economy-wide matter of importance.
So what’s the fastest solution?
The lowest latency comes from the most modern infrastructure – a true next generation network, ideally with full nationwide coverage that lets you transmit infinite amounts of data at the speed of light.
And yet not all companies that provide network infrastructure are the same. Many traditional carriers offer data networking effectively as a sideline, being more dedicated to the provision of commoditised telecoms voice services.
There are those who provide leased private networks, which offer performance and speed. But leased lines from a traditional telco are expensive.
There’s an argument for saying that network performance has progressed from being an IT concern to a boardroom one, and is now well on the way to becoming an issue of national importance. Only the right kind of next generation networks can provide the high capacity and low latency the UK economy will need to remain competitive for the future. If our enterprises don’t get the network speeds they need, they may end up suffering in the face of competition from rivals around the world. Network latency potentially affects us all.