3 ms·

I literally built billions of dollars of equipment for my employer deploying new datacenters. Even with a mandate to pivot to cloud-first, that kind of investm

I literally built billions of dollars of equipment for my employer deploying new datacenters. Even with a mandate to pivot to cloud-first, that kind of investment still needs to be used. And predictably after suffering enough pain from cloud-preferred thinking, we're settling on a mix, in no small part due to both monetary and migration cost.

candiddevmike 3y ago
Are you doing show/charge back for both on premise and cloud? Do different teams manage each environment?
- technofiend 3y ago
  >Are you doing show/charge back for both on premise and cloud? Yes, you're billed in either case. But it's the difference between blue and green dollars: money that exits the firm is much more strictly budgeted. > Do different teams manage each environment? Sort of. Knowing how to deploy to AWS vs an internally-hosted kube farm vs managing a fleet of virtual machines are all different skills, not that you can't learn them all. At one time our org structure split those functions but now if you go far up enough the management chain is the same.
  - dathinab 3y ago
    > deploy to AWS vs an internally-hosted kube This is IMHO one of the absurdities of current could technology. We have the technology and know-how to not make this different skills, to make it mostly uniform and abstract away all the compute hosting specific parts in the same way that compilers mostly abstract away most ISA/hardware details (at least for most languages and many use-cases, not all)
    - technofiend 3y ago
      I was disappointed (not that I'm picking on them) to learn Hashicorp's terraform didn't do exactly the sort of abstraction you're talking about. You can cobble together layers of automation, but I share your disappointment that there's not a single solution. Although I will say rackn's digital rebar gets a lot of the way there. You can deploy workloads in a pretty technology-agnostic fashion for the things it does support, and it can be integrated/extended with ansible, terraform and other scripting tools (including plain old bash) or other languages as needed.

candiddevmike 3y ago

Are you doing show/charge back for both on premise and cloud? Do different teams manage each environment?

technofiend 3y ago
>Are you doing show/charge back for both on premise and cloud? Yes, you're billed in either case. But it's the difference between blue and green dollars: money that exits the firm is much more strictly budgeted. > Do different teams manage each environment? Sort of. Knowing how to deploy to AWS vs an internally-hosted kube farm vs managing a fleet of virtual machines are all different skills, not that you can't learn them all. At one time our org structure split those functions but now if you go far up enough the management chain is the same.
- dathinab 3y ago
  > deploy to AWS vs an internally-hosted kube This is IMHO one of the absurdities of current could technology. We have the technology and know-how to not make this different skills, to make it mostly uniform and abstract away all the compute hosting specific parts in the same way that compilers mostly abstract away most ISA/hardware details (at least for most languages and many use-cases, not all)
  - technofiend 3y ago
    I was disappointed (not that I'm picking on them) to learn Hashicorp's terraform didn't do exactly the sort of abstraction you're talking about. You can cobble together layers of automation, but I share your disappointment that there's not a single solution. Although I will say rackn's digital rebar gets a lot of the way there. You can deploy workloads in a pretty technology-agnostic fashion for the things it does support, and it can be integrated/extended with ansible, terraform and other scripting tools (including plain old bash) or other languages as needed.