Raspberry Risk
Ever wondered how many Raspberry Pis it takes to run your risk analytics? Sounds like the setup for a joke, right? Its not. And we have an answer!
How many Raspberry Pis does it take to run your risk analytics?
Sounds like the opening line of a joke, and I’m sure someone funnier than I could probably write the punch line (please do suggest some below if you can!)
But as mad as it seems I genuinely thought it would be interesting to find out.
That was over a year ago. Finally, a week or two ago, I managed to get my hands on a Raspberry Pi 4B.
Didn’t take long to have it running Ubuntu Server and then spin up the COREx Financial Risk benchmark. Running COREx did take a while though! It’s not the fastest CPU in town.
Overall score? 30. Mean score per (logical) CPU? 7.5. For comparison, the lowest mean score per CPU out of all 650+ cloud VMs we tested was 14 (also an ARM based machine). That makes the Raspberry Pi twice as slow.
And, for comparison, a c6i.large has a mean CPU score of about 34.
Roughly speaking then, you’d need one Raspberry Pi 4B for every logical CPU in use. More if you’re using faster/newer machines. For a typical systemically important bank that’s somewhere in the region of 20,000 Raspberry Pis per region. Good luck trying to source them :)
Maybe next week I’ll compare the costs of running on a Pi 4B instead of cloud :)
