Feature Request: Storage Cluster / Storage Resource Scheduling

[ALindner]

Hello everyone,


first of all, thank you for your great work on Proxmox!
I know that DRS is high on your roadmap, and I’m very much looking forward to it – as I’m sure many others are as well.

In this context, I’d like to suggest a related feature:
A concept similar to “Storage Clusters” (as known from VMware), meaning resource scheduling and load balancing across datastores within a cluster.

Use Case:
Imagine a Proxmox cluster with multiple storage backends (e.g., several LVM volumes or other storage pools) attached as VM storage.
Currently, administrators need to manually decide where to place new VMs or disks. Over time this can lead to uneven utilization – some datastores filling up faster while others remain underutilized.


With storage-aware scheduling, the system could:

• Dynamically distribute new VMs/disks across available datastores based on current capacity and maybe even performance.
• Rebalance (storage migrate automaticly) when datastores approach critical utilization thresholds.
• Provide policies for preferred distribution (e.g., spread evenly vs. fill-up strategy).

This would ensure more efficient use of storage resources and reduce the risk of individual datastores becoming bottlenecks.


My question:
Should I open a new feature request for this, or would it be better to add it to the following existing report?
https://bugzilla.proxmox.com/show_bug.cgi?id=6243


Thanks a lot, and keep up the great work!

 

[dakralex]

Hi!

Thanks for the input!

My question:
Should I open a new feature request for this, or would it be better to add it to the following existing report?
https://bugzilla.proxmox.com/show_bug.cgi?id=6243

The linked feature request is still quite general in its description, but it is focused on the dynamic load balancing of HA resources first, i.e., the storage must be shared across the cluster nodes (or HA node affinity rules must be setup to ensure that a HA resource cannot be placed on a node that doesn't have access to that storage). There are plans to allow the resource scheduling on non-HA managed resources as well as pointed out on the roadmap [0], but that's quite a different feature request.

For the dynamic load scheduler, the current rough draft is to dynamically monitor the current CPU and memory usage/pressure and rebalance if there's a significant imbalance between the nodes that makes HA resource movement(s) justified. With that I could see for the foreseeable future that the dynamic load scheduler could also take account of the current I/O load of the node/guests and add that to the decision making, but take all of this with a grain of salt and as my current opinion.

So it might be better to put this in a separate feature request and make it depend on the linked Bugzilla entry [1], so one can see the relation between both.

[0] https://pve.proxmox.com/wiki/Roadmap#Roadmap
[1] https://bugzilla.proxmox.com/show_bug.cgi?id=6243