Related to the VM image project, we would like to see if it's possible to run the k8s operator-based W&B platform on Microk8s. If yes, we would like to understand the performance and scalability constraints of that, in terms of how large a single-node deployment could that sustain and what are its high-level breaking points. If it works for small (4 vCPU, 32 GB memory) to intermediate (32 vCPU, 256 GB memory) sized deployments, we would like to try and offer this as a v1 solution to customers who need capabilities unlocked via the operator but can't / don't want to use k8s directly, and see how far we can get with this.
Ask in this issue is to do a PoC on end to end deployment of k8s operator-based W&B platform on Microk8s, starting with local machine but then on a EC2 VM, and record the relevant instructions which could be shared with customers. When testing on AWS, try on at least three different sizes of VMs, and see how much memory Microk8s is able to provide to the application.
Related to the VM image project, we would like to see if it's possible to run the k8s operator-based W&B platform on Microk8s. If yes, we would like to understand the performance and scalability constraints of that, in terms of how large a single-node deployment could that sustain and what are its high-level breaking points. If it works for small (4 vCPU, 32 GB memory) to intermediate (32 vCPU, 256 GB memory) sized deployments, we would like to try and offer this as a v1 solution to customers who need capabilities unlocked via the operator but can't / don't want to use k8s directly, and see how far we can get with this.
Ask in this issue is to do a PoC on end to end deployment of k8s operator-based W&B platform on Microk8s, starting with local machine but then on a EC2 VM, and record the relevant instructions which could be shared with customers. When testing on AWS, try on at least three different sizes of VMs, and see how much memory Microk8s is able to provide to the application.