Earlier this year, Computer Weekly had the privilege of visiting JP Morgan’s development centre in Glasgow.
It was the first time the site had opened its doors to a publication. The centre is a hub of innovation, tapping into the rich vein of talent in the Scottish city and its universities. It houses teams dedicated to mobile technology, risk management and middleware development.
Much of the work is based on Java, but there are also many new and innovative uses of technology. In particular, some of the workloads JP Morgan is looking at push the limits of technologies such as Hadoop for real-time risk analytics. What makes this particularly interesting is that the bank has been a major user of open source technologies and it is now contributing code back to the open source community. This is a fundamental change in how the intellectual property associated with internal software development is viewed by business decision makers. In the past, commercial organisations considered software developed in-house as their own.
But, as Computer Weekly has previously reported, webscale organisations such as social media giant Facebook actively share code in the open source community. And now some of the world’s most closely guarded organisations are realising the benefits of publishing open source code.
When asked about the shift from creating proprietary software for internal use, to contributing to open source projects, Stephen Flaherty, who heads the developer team in Glasgow, told Computer Weekly: “We need a symbiotic relationship. There are a lot of challenges with the tools and we have done a lot of work to make sure everything is very robust from a cyber perspective.” In other words, collaborating and sharing code with the open source community helps the bank to make the software it develops internally more robust. And the open source community benefits from any code, documentation and problem-solving the bank’s developers share. The way an investment bank calculates risk may be highly guarded, but the JP Morgan case study in this issue illustrates that it is possible to share architectural details without giving away the keys to the safe. Perhaps one day its experiences of real-time risk analytics in Hadoop will be published as open source.