Replies: 2 comments
-
|
Management of the DigitalOcean cluster should be managed by Terraform/Tofu to support infrastructure as code that was started during PTS 2023. The Terraform provider is documented here: https://registry.terraform.io/providers/digitalocean/digitalocean/latest/docs/resources/kubernetes_cluster We should aim for smaller droplets with the ability to autoscale. Hopefully the autoscaling will allow us to resize the cluster as needed to conserve costs (more research required). |
Beta Was this translation helpful? Give feedback.
-
|
We have decided against using DigitalOcean for now - it would not be cost effective to run if we end up having to host elasticsearch ourselves. So I am going to close this discussion for now |
Beta Was this translation helpful? Give feedback.
-
DigitalOcean provides a relatively inexpensive managed kubernetes platform. They maintain the kubernetes controller, while we lease nodes from them to run our workloads. They are responsible for the redundancy and upgrading of the environment, we're responsible for the workloads required to run our applications.
This is only viable if we're not hosting our own ElasticSearch. The system requirements to run ElasticSearch are too costly to host in a managed environment like this, but also, on the same hardware that hosts ES, we'd have the capacity to host our own control plane (like we have at Hetzner now).
Beta Was this translation helpful? Give feedback.
All reactions