> operationally simpler and easy to recover from instance failures than scaling per se.
I'd say architecturally making the app easier to perform this function, rather than rely on infrastructure to do it for you (and hoping doing so will make the app "easier" to code).
For example, use an append only data structure (like event sourcing), with snapshotting. Then your app recovery process is just "restart".
I'd say architecturally making the app easier to perform this function, rather than rely on infrastructure to do it for you (and hoping doing so will make the app "easier" to code).
For example, use an append only data structure (like event sourcing), with snapshotting. Then your app recovery process is just "restart".