by HenryBemis on 6/20/2025, 3:43:10 PM
It is there, it is online, it can scale 'in seconds', it can be reached by many (if necessary) and it is easy to update (with great care and testing so it doesn't break 50% of the customers business/game/etc).
So, why not use it? If one has the appropriate ITGCs and a decent risk assessment, then go for it!
by v5v3 on 6/20/2025, 9:15:12 AM
Were the engineers incentivised and measured on minimising cloud services use?
As the end customer, be it the internal or external, will give a set of requirements, and as long as the solution meets those requirements the project has been delivered.
Two examples:
I recently saw an engineer provision an S3 bucket to store a temporary file that is only needed for the current session, and only accessible to the application that created it. For me, this is what the local file system is for.
I saw another engineer provision an in-memory key-value store to store some infrequently changing reference data that can be easily loaded at startup, in an application runtime where memory persists between requests. For me, this is what local heap memory is for. I would simply load the data into a singleton map.
Both of these services have legitimate use cases, but I can't help but wonder whether they (and other tools and services I won't go into here) are beginning to make programmers forget that our applications run on machines with a standard set of resources (CPU, memory, disk, network interface) made available through an operating system, and further abstracted by the API provided by the development platform.
Has anyone else experienced this?
[1] The systems I worked on were non-trivial and mission critical: capital market trading systems. So it's not a case of those "dumb" solutions being either non-resilient, non-performant or unmaintainable.