[axg11]

How else would you scale to meet peak demand without being wasteful?

Banking has a fairly predictable usage pattern, but there will be black swan financial events that cause 100-1000x load. On top of that, how else could you serve customers around the globe with reasonable latencies?

These are genuine questions. I’ll admit I’m an engineer whose entire career has been during the cloud era. I don’t see how cloud’s advantages of scaling and worldwide “edge” locations can be replicated by the average bank’s tech team.

[toomuchtodo]

> How else would you scale to meet peak demand without being wasteful?

I can only speak to the banks I've have the opportunity to work with who stayed on prem or built their own internal cloud infra; what you call waste, they consider a premium/cost of business for security and resilience. I'm sure a lot of folks would've said the same thing about JIT supply chains (that had been squeezed to be as efficient as possible) until they unraveled.

> I don’t see how cloud’s advantages of scaling and worldwide “edge” locations can be replicated by the average bank’s tech team.

As always, "what are your requirements and what are you optimizing for?" Most folks don't need web scale nor edge locations [1] [2], they'll get by just fine with a CDN and some API endpoints [3].

[1] https://news.ycombinator.com/item?id=19576092

[2] https://blog.bradfieldcs.com/you-are-not-google-84912cf44afb

[3] http://mcfunley.com/choose-boring-technology

[axg11]

Are there many examples of companies (today) building 1000x the infrastructure they normally need? I can see how it could be necessary for some companies.

> Most folks don't need web scale nor edge locations [1] [2], they'll get by just fine with a CDN and some API endpoints [3]

Isn’t using a CDN using cloud?