Scale Computing VS Proxmox
-
Hi Everyone,
We have been using Scale Computing for the last 5 years and are up for a full stack replacement now. I was wondering if anyone has any opinions on moving from Scale to something like ProxMox? I would be very appreciative to hear any advice/opinions on this or other solutions. Just want to make sure I make the right decision for our company moving forward. Our server load is not super intense, it is 9 MS Terminal servers, 2 Exchange Servers, 2 AD servers, a couple of file share servers(both MS and Ubuntu), and then a unifi controller and that is about it. I should also note that we are also looking to change our Backup and DR vendor as well as we have been using Unitrends but have been unhappy with them since the Kaseya acquisition. Just in case that makes a difference. Thank you everyone in advance! -
@NHCSAdmin said in Scale Computing VS Proxmox:
I was wondering if anyone has any opinions on moving from Scale to something like ProxMox?
These two stacks serve different purposes.
Scale is Hyperconvergence
Proxmox is simple virtualizationYes there is a lot of overlap between the two depending on how things are setup, but you should ignore trying to squeeze something into a non-standard shape, even if it can fit.
-
@NHCSAdmin said in Scale Computing VS Proxmox:
Our server load is not super intense, it is 9 MS Terminal servers, 2 Exchange Servers, 2 AD servers, a couple of file share servers
Actually, it is pretty intense. Two exchange server and 9 Terminal Servers? That is not a light workload. Sure, it is not huge Enterprise, but that is well beyond basic SMB needs.
-
@NHCSAdmin said in Scale Computing VS Proxmox:
I would be very appreciative to hear any advice/opinions on this
So, both solutions could work for your need.
How are you currently using the Scale nodes? Are you making use of the benefits of the hyperconverged architecture?
If not, then it would be pretty simple to setup everything on a few physical servers in a Proxmox cluster.
Proxmox is going to be more manual work to move things around during updates compared to a Scale system.
Hyperconvergence can basically just be considered as virtualization with all the pieces in a single pane of glass and completely seemless control. The details are more complicated of course, but that is a workable analogy.
-
@JaredBusch Thank you so much for the replies! For the most part we are using Scale's benefits over the simple virtualization only when there is a Scale HC update. The rest of the time our VMs do not really move at all between nodes. We have been happy performance wise with Scale but I wanted to make sure I explored options prior to just moving forward with a renewal. As a team of 1 I wanted to make sure I wasn't missing anything obvious by just being swamped all of the time if you know what I mean. I really appreciate you taking the time to reply and explain.
-
@NHCSAdmin said in Scale Computing VS Proxmox:
@JaredBusch Thank you so much for the replies! For the most part we are using Scale's benefits over the simple virtualization only when there is a Scale HC update. The rest of the time our VMs do not really move at all between nodes. We have been happy performance wise with Scale but I wanted to make sure I explored options prior to just moving forward with a renewal. As a team of 1 I wanted to make sure I wasn't missing anything obvious by just being swamped all of the time if you know what I mean. I really appreciate you taking the time to reply and explain.
Scale and other HC platforms are really designed around simplicity from the IT department perspective and having "system level" failover. Meaning it's not as robust as application layer failure, it's the standard fallback for shops that either don't have the full stack skillset to do application layer failover and / or are stuck running applications that don't support it. If that's where you are, this is going to be the best choice for you. If you have application layer failover in your apps, then HC doesn't make sense as it is protection that serves no purpose.
Most shops have some amount of hands being tied so system level HA is a big deal for them.
-
@scottalanmiller thank you! We definitely have some apps that some of our users are using that do not offer the ability for application layer failover. Of course those are the apps that the company uses for billing and payroll so they are critical. Based on your's and @JaredBusch's responses it definitely seems like it is worth looking into the renewal of the Scale cluster.
-
@NHCSAdmin said in Scale Computing VS Proxmox:
@scottalanmiller thank you! We definitely have some apps that some of our users are using that do not offer the ability for application layer failover. Of course those are the apps that the company uses for billing and payroll so they are critical. Based on your's and @JaredBusch's responses it definitely seems like it is worth looking into the renewal of the Scale cluster.
Yup, makes sense. of course non-HC makes sense where you need no failover at all, as well.
Your AD controllers are the perfect example of application layer failover. They have that built in so you don't want them being moved by the virtualization layer.
Your accounting needs failover, but has none of its own. So needs the crutch of the HC system.
Your Unifi controller has no problem with downtime, so no failover is necessary at all. Since you have it, might as well use it, but if the controller went down for a while, no big deal.
All three workload scenarios there. Since you have the middle one, you need HC or similar.
-
@scottalanmiller Thanks for the detailed explanation!
-
@NHCSAdmin said in Scale Computing VS Proxmox:
@scottalanmiller Thanks for the detailed explanation!
You bet! Glad to assist.
-
Did scale ever add the ability to unmount and remount disks? The last time I used it (2018) you couldn’t unmount a disk and then mount it to a new machine.
-
@stacksofplates said in Scale Computing VS Proxmox:
Did scale ever add the ability to unmount and remount disks? The last time I used it (2018) you couldn’t unmount a disk and then mount it to a new machine.
They did but I don't know the process.
-
@stacksofplates Yes, we did in 2019. There are a couple of ways that can be done. When you snapshot a VM, any disk in that snap can be mounted to any other vm, provided that the logged in user is at a permissions level allowing it. That is actually part of the mechanism that several backup vendors (acronis, storware,etc) use to do agentless backups of Scale Computing VM's. If you haven't taken a look since 2018, you should take a look again as there has been so very many things added since then.
-
@Aconboy said in Scale Computing VS Proxmox:
@stacksofplates Yes, we did in 2019. There are a couple of ways that can be done. When you snapshot a VM, any disk in that snap can be mounted to any other vm, provided that the logged in user is at a permissions level allowing it. That is actually part of the mechanism that several backup vendors (acronis, storware,etc) use to do agentless backups of Scale Computing VM's. If you haven't taken a look since 2018, you should take a look again as there has been so very many things added since then.
A wild @Aconboy appears!
-
@scottalanmiller I gotta show up from time to time or people think I am a myth
-
@Aconboy said in Scale Computing VS Proxmox:
@stacksofplates Yes, we did in 2019. There are a couple of ways that can be done. When you snapshot a VM, any disk in that snap can be mounted to any other vm, provided that the logged in user is at a permissions level allowing it. That is actually part of the mechanism that several backup vendors (acronis, storware,etc) use to do agentless backups of Scale Computing VM's. If you haven't taken a look since 2018, you should take a look again as there has been so very many things added since then.
Well I guess I mean more without doing a snapshot. The flow we were looking for at the time was we had an ephemeral VM that would boot, mount the disk for storage, we could unmount the disk, destroy the VM, bring it up a new copy and remount the disk. The data disk wouldn't be in the snapshot since it only holds app state. Think of it like persistent volumes for k8s.
