I live in the countryside. As a city boy myself, this was everything I was against. Gone is the excitement, the restaurants, the bars, the constant drone of traffic, the smells to be replaced with sleepy small town life, the sound of cows and birds and the smell of cut grass. My younger self would be appalled.
There's one thing I didn't count on when I moved though... Tractors. You see, in the city, I had a pretty predictable route to work with pretty predictable traffic on large open motorways. I could safely bank on a consistent twenty minute commute to the office and back, regardless of time.
Now... In the country, I've got tractors. These trundling behemoths can turn a short journey into an epic. On twisting country roads, no opportunity to get past, just a long slowly moving queue of traffic. It's no longer safe for me to assume a quick drive to the train station. It could be ten minutes, it could be half an hour. The vast majority of the time, it's quick and painless... But every now and again, just to keep me on my toes, the tractor appears and slows me down considerably.
Had I not compensated, I could miss my train, miss my plane, miss my meeting with a client and that would have a direct impact on my business.
In the world of IT we'll measure our round trip latency, quote our average number, and then just assume or hope that it'll be consistent. As we all know, this is never the case. When things are going smoothly, when the sun is shining, the birds are tweeting (in chirps of 140 characters or less) and the IT systems are happily humming away in their datacentres then our latency is consistent. From a tick to a quote in our FX trade system, we're consistently knocking it out the park.
But things go wrong...Latency isn't a single number. It’s a distribution. There are numerous complex components that contribute to the latency of a trade. It's not just the network, there are the applications and the supporting infrastructure, the 3rd party components and the laws of physics. Every now and again, the IT Tractor will pull out and drag that latency down.
This is incredibly hard to predict, and incredibly difficult to track down. While I can compensate for my longer trip to the office, this simply isn't possible in a highly competitive market place. So the 'long tails' need to be dealt with. They may be relatively rare, but when these spikes in latency occur, they have a real impact on business.
To deal with this requires a holistic view of the capacity, performance and latency across the entire IT stack. A real understanding of what components contribute to the latency, how their contribution differs over time and what circumstances cause that contribution to increase. How does that increase in latency correlate with volumes and resources. When we have this model in place, we can really begin to drill down and target those problems in the system and start to move towards a more consistent and lower latency distribution.
Now... If only I could do something about that tractor.
Posted by Scott Russell, Sumerian Product Manager
In the 70’s people wore their trousers flared, ate prawn cocktail and lettuce from glass bowls and carried their music around on 12 inch vinyl. 20 years later, trousers had holes instead of knees, Vienetta was on the menu and everyone was wiping jam off their indestructible CDs. Kids saved up their pocket money to buy 3 minute singles from high street music stores and most people were happy to pay £12 for an album that they had to get a bus into town to pick up.
Here at Sumerian, we’re loving the World Cup. This is despite (or for some people, because of) England’s poor performance. We’re running a Dream Team competition and organised a Sweepstake which pays out if your team reaches the semi-final. But these things are notoriously hard to predict – spare a thought for Alastair, who drew Brazil and must have thought he was a dead cert to win something, or Roly, who I personally assured when he drew them that Italy stood a great chance of going far in the competition. And who would have predicted a Uruguay-Holland semi-final even on Friday morning?
It’s more important than ever for organisations to have a strong handle on their IT capacity. Business volumes and technology are changing at an increasing pace – more and more IT Services rely on virtualised infrastructure with configurable and shared capacities, and IT shared services (or outsourced services) are costed and priced on the capacity provided.
On the WAN, you might be looking at milliseconds or tens of milliseconds, depending on the distances covered. That’s orders of magnitude larger than the LAN, so you might immediately think it’s worth focusing on. The problem is that in this area your potential improvements are limited by another of Einstein’s observations – nothing can go faster than the speed of light. The speed of light in a fibre (fiber for any American readers :-)) is about 2/3 of its speed in a vacuum, so in the region of 200,000 kilometres per second. This means that every hundred kilometres you have to travel is going to add at least half a millisecond to your latency. Need to get data from New York to Chicago? (roughly 1100 km/700 miles). Not going to happen in less than five and a half milliseconds no matter what you do (and probably more than double that because fibres, unlike hypothetical crows, don’t follow straight lines).
With the recent weather showing no signs of letting up in the next few weeks in terms of temperature, and with further snow storms forecast, various seams in the UK Infrastructure are beginning to show signs of stress.
Cloud computing is the "next big thing" in IT, or so the media would have us believe. The reality is that cloud computing has been around for a while, it is however only recently that the “cloud” has started to get mainstream media attention as the big players like Google, IBM, and Microsoft are pushing their “cloud” solutions. You may be more familiar with the term utility computing. Whatever you call it, the shift to the cloud is gathering momentum. We should understand it, embrace it, and where possible replicate it.
It’s fair to say that, here at Sumerian, we’re generally pretty fond of numbers. These range from the sublime, to the ridiculous.
The design requirement was therefore extraordinarily demanding. To meet it, the Apollo programme produced a spectacular device. Based on ICs, the 
