Building a Reliable Cloud-Based Bank in Java

[@clray123]

In short: record any request that you accepted; never drop the ball on processing yet unprocessed requests; make sure that every request is processed only once. Sounds obvious, but you'd be surprised how many people screw it up. The usual way of exception handling is "log it then let the world collapse".

[@mihaiciureanu923]

Very nice presentation, but it kind of reinvents the wheel. We already have bounded contexts, and sagas. Also, it would have been nice to hear what happens if something can't be processed somewhere down the stream , which is often the case of a bank. For example there are not enough funds at the exact moment where the message gets to the ledger. Normally there are compensating transactions in this phase, and I would have liked to see how they are implemented.

[@EvgeniiSuvorov]

bblablabla where is the code?)

[@-rpm]

This is what happens when you hire incompetent engineers. No matter how many code reviews, roll back plans you have if it is being done by incompetent engineers it will fail terribly. One month for a roll back in a banking infra is ridiculous.