4 ms·
This is partly true. The reason AWS et al ever grew in the first place was also that managing data centers well and automatically used to be a lot harder, and w
by andrewstuart2 3y ago
This is partly true. The reason AWS et al ever grew in the first place was also that managing data centers well and automatically used to be a lot harder, and was assumed to be a whole org's full-time mostly manual job. AWS, Google, et al showed how well this could be done with software, showed the benefits of making it easier to do even for app engineers, and incentives shifted. Tools have significantly improved, though, and we're getting back to the point where it can be done with OSS and a team or three of engineers who really know the stacks involved.
- Nextgrid 3y agoIn my opinion, the promised "simplicity" of the cloud never materialized. You've replaced physical hardware that's easy to reason about with opaque, unpleasant and eventually-consistent APIs along unnecessary complexity (VPCs, etc) only designed to vendor-lock you even more into the provider. The staff savings never materialized - every company I've been involved with had at least one "DevOps" engineer, but instead of fighting with hardware he's constantly messing with YAML/Terraform/CDK files.
- politelemon 3y agoThe alternative isn't that different however. A company running its own centre may choose to go OpenStack/k8s. It still requires lots of messing about with awful files. The 'cost' then shifts elsewhere, and merely satisfies bean counters. IMO the simplicity promised does exist, it's up to the company how it gets implemented. Like either toolstack, if it's misused, it will cost more than they thought (and then they'll blame the toolstack)
- SoftTalker 3y agoI've tried several times and never got an on-prem k8s working. I use proxmox now and never looked back.