In the last couple of years, investment banks have been having quite a tough time, things have not been very easy for them, although now it seems there is something of a recovery taking place.
One of the ways they have been able to get through these difficult times is through more intelligent use of technology, to help them save on costs, for example.
However, the banks will need to continue looking at how they can make best use of technology if they want to stay compliant with the various regulations that are coming out around risk management procedures and practices.
In order to perform the required risk modelling in this new world order, it is estimated the banks will need something like 10x the computing power they were using before the crisis. So the question is, where will they get these resources? Existing data centres are becoming heavily over-utilized and there is little spare capacity. In fact, there is one investment bank reputed to use more data across its data centres than the entire city centre of Manchester! This may be an urban myth but it's probably not too far from the truth.
Accordingly, many banks are now looking seriously into making more use of cloud computing and grid computing services. This could be a good option, particularly because of the way banks utilise computing power. Typically, in order to run risk simulations, they need to make very heavy use of computing at certain times of the day, whereas at other times, their computers are virtually idle. So utilization of existing resources probably runs at less than 50% some of the time.
Sharing compute resources and using cloud computing would seem like the sensible answer, but there are problems associated with that approach. First of all, the high performance computing (HPC) applications typically run by investment banks don't sit too well on the cloud. This isn't the biggest issue however. The key problem is that banks generally don't like to share, particularly not with their competitors. Security is also a big issue, although cloud services are generally seen as pretty secure these days, security issues around cloud are still causing reluctance on the part of the banks. Then of course there is the whole issue of Service Level Agreements (SLAs), for example the need to complete risk calculations by a specific time of the morning, which is another problematic area with shared resources.
The answer could be a new paradigm known as Grid as a Service (GaaS), whereby high performance compute is offered on-demand, across the grid, on a pay-as-you-go basis. Banks could encapsulate specific calculations and send them off to allocated compute resources in a shared environment. This would allow banks to exploit all the various cloud infrastructure, services, middleware and compute power.
GaaS is certainly set to grow and could change the face of computing in the investment banking world.
