Thinking about replacing VMWare with Proxmox - thin provisioning question

ATN Chris

New Member
Dec 23, 2022
2
0
1
Hello everyone! I have recently learned about proxmox and started researching to find out if it's a good fit for us and what we would need to replace our current infrastructure. We currently have 2 separate ESXi clusters with a mix match of hosts (3 identical HP hosts, 4 identical linux hosts) and iscsi SANs (2x EMC vnxe 3200 and 1x EMC Unity 300). My concern is I've been reading that thin provisioning is not possible with iscsi SANs. I wanted to first, confirm this - and second, if it is not possible is this something that would be possible in the near future?
 
thin provisioning is not possible with iscsi SANs
There are at least two layers where thin provisioning can happen. First is inside your SAN, that will stay available regardless of the hypervisor.
Second is at the hypervisor layer during virtual disk provisioning (disk image in Proxmox terminology). There is no VMFS equivalent built-in with Proxmox. You have a few options:

- use LVM thick on top of your iSCSI LUN. As the name implies the LVM slice allocated for each virtual disk will have all blocks reserved from start. There is also no snapshot support.
- Create a SAN LUN per disk. This approach requires a lot of manual management overhead as there is no integration for Dell/EMC SAN and PVE. This might buy you some thin provisioning, since you will be relying on the backend storage to no allocate everything.
- Install, configure and manage one of the publicly available Cluster Aware File systems that provides the functionality you need. The management and integration of it will be completely on you.
- Replace your aging EoL SAN with a modern solution that is PVE integrated.

is this something that would be possible in the near future?
IMHO, I'd say it is unlikely that such functionality will be implemented in near future. You are welcome to file a feature request so that PVE team is aware of your desire.

P.S. you may find this interesting if you are building a business case for hypervisor change:
https://kb.blockbridge.com/technote/proxmox-vs-vmware-nvmetcp/


Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Last edited:
- use LVM thick on top of your iSCSI LUN. As the name implies the LVM slice allocated for each virtual disk will have all blocks reserved from start. There is also no snapshot support.
The lack of snapshot capability is not necessarily a deal breaker for me as long as Proxmox Backup can still image the VM. (I have zero knowledge on this product). If Proxmox Backup uses PVE's snapshots to do it's backups then I suppose that will be a deal breaker.

- Install, configure and manage one of the publicly available Cluster Aware File systems that provides the functionality you need. The management and integration of it will be completely on you.
I wouldn't know where to begin here but it sounds like it might be promising if it allows snapshots and thin provisioning. Would you happen to know an example of such cluster aware file systems? Would Ceph be one of those as that seems to be mentioned as a big one for the hyperconverged storage.

Speaking of, If I wanted to just add local storage to all my nodes to take advantage of hyperconverged storage, would that be better than purchasing a new SAN? The costs would probably be comparable.
 
If Proxmox Backup uses PVE's snapshots to do it's backups then I suppose that will be a deal breaker.
PBS does not take advantage of the storage snapshot ability. It utilizes QEMU dirty bit tracking.
I wouldn't know where to begin here but it sounds like it might be promising if it allows snapshots and thin provisioning.
Probably something you should be familiar with before you place production workload on it. Also keep in mind the "Bus Factor" and the fact that this will not be supported by PVE team regardless of your subscription.
Would you happen to know an example of such cluster aware file systems?
https://wiki.samba.org/index.php/Setting_up_a_cluster_filesystem
https://en.wikipedia.org/wiki/Clustered_file_system#Examples_2
Would Ceph be one of those as that seems to be mentioned as a big one for the hyperconverged storage.
In a way Ceph would be part of the list. However your storage is not hyperconverged now, the SAN is external/centralized. Using Ceph with SAN defeats the purpose of both.
Speaking of, If I wanted to just add local storage to all my nodes to take advantage of hyperconverged storage, would that be better than purchasing a new SAN? The costs would probably be comparable.
If your hosts and network are as old as your SAN, personally, I would be weary of trying to invest in them. Hyperconverged means that the host is running both Compute and Storage (+network replication). Are your hosts' CPU up to it? Do you have enough RAM?
There are many documents that specify minimal suggested requirements for Ceph, i.e. https://pve.proxmox.com/wiki/Deploy_Hyper-Converged_Ceph_Cluster .
Yes, people are running Ceph on a mix of home computers with HDD and 1G network. It all depends on your expectations and requirements.

Only you have all the facts to know what the right investment to make.


Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!