1) MISYS has an office in Gdynia, Poland that develops financial software.
2) The office was originally created in 2006 as part of Thomson Reuters and became part of MISYS in 2012.
3) The Gdynia office currently has around 100 employees working on development, quality assurance, and support of MISYS products like Kondor+ and Misys Global Risk.
[...] I work in Gdynia with a team on the Liquidity part of MGR. I’d like to speak about performance First let’s start with the performance of our products – just as an introduction.Why is it needed by our customers?This is something that was implicit in the movie you saw earlier. If you’re not familiar with the financial industry, it may appear normal, but it is very common for VaR to take 1 night to be calculated: it’s normal there are billions of calculations required to calculate this.So to be able to measure risk with rapid changes in the market is a huge advantage we give our customers.We not only need to be fast, but we need to be scalable, because the number of trades are increasing: more and more use of automated trading.How do we do it?We have a scalable architecture: by dispatching the calculations to several machines, and using a distributed cache, you give better performance.The next bottleneck becomes the storage of these calculations, so we use a no-SQL storage, a bit like Hadoop; something custom developed in-house; in other areas of the product we’re using GPUs for pricing
Now, I’m a developer. And if you looked at our ‘performance’ not so long ago... it was not brilliant.Why?Well, we’re working on a big project. 2.3 million lines of java.We were using clearcase for source control, and while some might argue it is a good; well, (I would disagree with that), it’s slow.Maven2, do you know cannot do parallel buildsI think there are some developers in the audience, so you know what I’m talking about. This is common on large projects, and it’s frustrating, and this was depicted by the famous Randall Munroe on xkcd [read some of the drawing]Yeah. But in fact, it’s not really funny, it’s more frustrating. We like to code, but we have to wait for our computer to let us code!Also, we had some tests, but it took too long to execute them too, and it was up to the developers to run them, and the results were not published, and it was not done frequently enough... You know, the continuous integration story.So... There was an opportunity for improvement. And we found a recipe:
Throw in some hardware:Very fast SSD drives (the key indicator for us was random writes), Add 32GB of RAMSome software: linux, Git very fast and resilient, makes working with different locations a breeze,maven3 builds in paralleland jenkins, we have 8 nodes that compile and run 3 different types of test on every commit.... And a large screen
plus 2 other screens.Show a jenkins feedback sceen to everybodyAnd there you go. We now build in less than 5 minutes. Running tests is simple, and everybody can see the results, and who broke them,We work comfortably and we’re back to take pleasure in coding. And when we take pleasure at what we do, we do an excellent job.And when we do an excellent job, our customers like it, and so does managmenent.So they clearly see the return on investment.And I would like to point out that many of these innovations like git, maven3, and a better use of jenkins came from Gdynia