3 ms·
These are completely different scales though. If your systems can be run on a rack of servers in a closet at the office by a part-time sysadmin, you're probabl
by cassianoleal 4y ago
These are completely different scales though.
If your systems can be run on a rack of servers in a closet at the office by a part-time sysadmin, you're probably better off doing that instead of running said k8s cluster.
If you do need the scale of said k8s cluster, pipelines, operators, observability, on-call engineers etc, you're better off doing that but it does come at a cost.
- hedora 4y agoThe vast majority of the jobs are at the small end of the scale. Also, companies tend to just use best practices that were established for larger scale operations. For instance, in its heyday, the vast majority of Hadoop installations had under ten machines. I suspect the same is true of spark and kubernetes.