Differences in FSI HPC Architectures
Most financial risk system looks more or less the same? What about when they don’t? You might think there are great fundamental reasons for these. Good reasons for why these variances exist. But you’d be wrong.
Often high performance computing architectures in FSI, through a process of convergent evolution. look broadly similar even if they arrived at the result from mostly different starting points,. (See this blog by Alex Kimber for more on convergent evolution). What is more interesting than the similarities though, are the differences. Are there good reasons for the divergence?
I speak largely from experience with HPC architectures in financial services, but I suspect the same is true for many large enterprises.
Enterprise software systems are expensive. They are also long lived. In a typical enterprise that isn’t styling itself as tech, or more likely these days, AI company this means they’re a spendy cost centre and treated as such. If this sounds like a recipe for creating “legacy” systems quickly that’s because it is. Unfortunately, financial risk systems tend to quickly fall under this classification.
More often than not, they are also a reflection of the internal organisation and politics of the enterprise. And financial risk systems by their very nature are required to interact with a myriad of other enterprise systems to source and deliver data.
The result? Probably the most operationally expensive system has, to some degree, its architecture constrained by a plethora of other far less complex systems (though potentially equally important).
Oh and then you have migrate this meatball minefield laden bowl of spaghetti to the cloud. Without upsetting CISO. 😂
