Sheogorath on Nostr: npub1u2hl9…4z4d7 well, quite wonderful article with mainly trying to sell some AI ...
npub1u2hl9r48emg8eq82nh8zwadl3xr93u7vkk37qavfcg7cu3dmrrzqe4z4d7 (npub1u2h…z4d7) well, quite wonderful article with mainly trying to sell some AI magic.
Reality is, for most use-cases CPU is highly volatile and needs to be available in spare in order to achieve ideal performance for applications.
This means horizontally scaled workloads become less and less efficient. Having CPU usage between 10-20% utilisation is completely normal for most workloads.
This is also independent of Kubernetes. Same goes for Hypervisor clusters and more.
Reality is, for most use-cases CPU is highly volatile and needs to be available in spare in order to achieve ideal performance for applications.
This means horizontally scaled workloads become less and less efficient. Having CPU usage between 10-20% utilisation is completely normal for most workloads.
This is also independent of Kubernetes. Same goes for Hypervisor clusters and more.