Moving off VMware Hypervisor to something else - need input
-
@scottalanmiller said in Moving off VMware Hypervisor to something else - need input:
@dave247 said in Moving off VMware Hypervisor to something else - need input:
Looking at Nutanix VS Proxmox and surface level perception is that Proxmox could be better
I would quite literally place one as the worst possible option and one as the best. VMware would be close to Nutanix on the bad side. Nutanix, VMware and Hyper-V these days I don't know any argument for how to use them in production.
I like ProxMox and XCP-ng for pricing, licensing, experience, etc. There are loads of KVM systems out there, ProxMox is far from the only player. But I think you'd need a pretty incredible argument to use one given that most or all will be closed source (that means "not production ready" to us) and few have the experience, track record, support, industry knowledge and internal support of ProxMox.
Thanks for the input Scott. All I have really used for the last 10 years is VMware so I just did a surface level search for other Hypervisors, so I'm not really even considering Nutanix, it was just one that came up. Anyway, everything I'm reading suggests that Proxmox is rock solid and that sounds like what I'll probably test out for a while as a VMware replacement. Plus it's free which is a plus, especially since VMware renewal would be insanely costly. I can check out XCP-ng too.
I really just want something that is stable and easy to work with. We don't have that complex of an environment so the only things I'd be doing is server firmware upgrades and Proxmox updates when needed. Doing that stuff with VMware always made me nervous since something would often break.
-
@travisdh1 said in Moving off VMware Hypervisor to something else - need input:
I'm in agreement with Scott here. There is a very short list of options, and Nutantix is not one of them.
Proxmox would be the primary choice (the backup server is really easy to work with as well), and XCP-NG if Proxmox can't be used.
Migrating from a VMWare to Proxmox is also really easy. I did a trial at a former work place.
Thanks. That's great to hear that migrating from VMware to Proxmox is easy. And I'm currently using Veeam for our backups and that works well with Proxmox from what I'm reading.
-
@DustinB3403 said in Moving off VMware Hypervisor to something else - need input:
@dave247 Going from expensive VMWare to expensive Nutanix makes little sense.
Yeah I was just googling other Hypervisors and looking into each. Proxmox sounds like a really good option and I'm going to test it out in my lab next week.
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
I can check out XCP-ng too.
You should. It is fully open source, so the only thing we (Vates) are selling is professional support and maintenance. LMK if you have any questions about it.
-
@dave247 ProxMox has its own built in backups, too. Not as powerful as Veeam, but enough for many environments. We have a LOT of ProxMox out there and none use any third party backup.
-
FoxRMM is working on ProxMox backup monitoring being centralized and included in its next release too.
-
@scottalanmiller said in Moving off VMware Hypervisor to something else - need input:
FoxRMM is working on ProxMox backup monitoring being centralized and included in its next release too.
When does the rest of the world get a look at FoxRMM?
-
@scottalanmiller just out of curiosity, could you provide any arguments against using Hyper-V?
We are 99.9% Windows PC & Server shop where I work so naturally some might suggest us using Microsoft's Hyper-V. I have used it a handful of times in the past but it didn't seem very user friendly and seemed to have issues at the time, granted it was over 8 years ago.
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
@scottalanmiller just out of curiosity, could you provide any arguments against using Hyper-V?
We are 99.9% Windows PC & Server shop where I work so naturally some might suggest us using Microsoft's Hyper-V. I have used it a handful of times in the past but it didn't seem very user friendly and seemed to have issues at the time, granted it was over 8 years ago.
Sure...
-
Microsoft no longer has faith in it, it would border on insane to trust as your core platform something that even its own vendor gave up on. The only production style deployment was removed and now only the dev / hobby style is still available. People argue "they didn't discontinue it" and no, not technically, but they did discontinue it from anything that could be argued to be a production methodology. Using HyperV means defining your virtualization infrastructure as a "hobby" environment where uptime and reliability don't matter.
-
It doesn't have the knowledge, development or care of a production system. Never did, but now its a tiny fraction of what it once was. So if you thought it was bad back when MS still pushed it, image how bad it is now!
-
No support. It's an MS product, that means you are left to your own devices and the third party support world is essentially entirely resellers, not skilled engineers. No serious support engineer works with HyperV, so you are depending on the market of somewhat technical salesmen to be your support network.
-
Licensing. Pure and simple, if your infrastructure product requires a license, it's a joke. Nothing causing outages like a license. That's why no serious business will touch that stuff. Risk for no reason is literally a form of sabotage.
-
Confusion. While someone, somewhere installed HyperV and knew what it was, you'll never find that person. Everyone who recommends and implements it does so out of thinking it is something that it is not. Many swear it is a Type2 (and then think that that is a good thing.) Almost no one can actually identify what is virtual and what is not when using it. Especially if you have a Windows show, the chances that anyone really will know HyperV is zero. Only very senior UNIX shops ever have the knowledge to really understand it and they avoid it based on that knowledge.
-
Cost. HyperV isn't free anymore and while someone might argue that you are SO tied to Windows that you'll never be able to make decisions based on what's best for your business, that's a horrific financial mentality to take. Again, risk without benefit, even if small, is very bad. It's what IT pros should be paid to protect against, not recommend. It's a product that only works against your interest, not for them.
-
Management. Like most Windows products, Hyperv's management is a sh1tshow and a completely joke compared to ProxMox and XCP-ng as examples. Not even in the game. If you thought VMware was bad, HyperV is the only thing that gives it serious competition for being so bad.
-
Source and Security. HyperV is closed source and from a notoriously unstable, insecure vendor. You might be stuck with Windows servers, but there's no reason to extend those problems into other arenas without benefit. One mistake does not justify another. Saying "we didn't care about price, security, auditability, performance, stability, etc. with our servers, why should we care now" is the antithesis of our jobs. It's worth noting that all those things clearly aren't top priority, so there might be other factors that we can weigh more heavily. But HyperV lacks ANY benefit (not even familiarity or ease of use!) to Windows admins (ProxMox is literally easier and more straightforward out of the box) so even the seemingly obvious factors work against it. So while giving low weight to production factors is okay, lacking other factors to outweight them doesn't change the outcome.
-
You'd never be able to feel proud of your environment or tell anyone what you run on without feeling embarrassed.
-
Backups, remote support, remote monitoring, etc. all harder. ProxMox and XCP-ng shine at how easy they are to automate, to manually manage, to manage from anywhere. HyperV falls flat in all those areas.
-
THis is a cheat benefit (IT should NOT consider this as it is cheating the company) but HyperV is a valueless skill. You don't want to have that on your resume instead of something of value. So every single person in your IT department should be up in arms at the idea because nothing about it makes their lives easier, and nothing about it is good for their personal careers. Everyone loses, no winners. Again... as IT pros, we should be taking the value to the business first and not making decisions based on what is good for us personally. But this is a care where those two things align.
-
Performance. Haven't checked this recently as HyperV is abandoned in production so it doesn't matter, but historically KVM was the performance leader for Windows workloads.
Most importantly, I can't think of any reason someone would give for even considering HyperV. The utter lack of upsides, being the sole product that EXISTS that's actually worse than, rather than better than, VMware is really something.
-
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
@scottalanmiller just out of curiosity, could you provide any arguments against using Hyper-V?
We are 99.9% Windows PC & Server shop where I work so naturally some might suggest us using Microsoft's Hyper-V. I have used it a handful of times in the past but it didn't seem very user friendly and seemed to have issues at the time, granted it was over 8 years ago.
- Hyper-V standalone is being depreciated. It will cost you a Windows Server license to continue running.
- Functionally more limited than any other option.
- Performance.
There are reasons why not even Microsoft runs the entirety of their cloud services on their own platform.
-
@scottalanmiller said in Moving off VMware Hypervisor to something else - need input:
@dave247 said in Moving off VMware Hypervisor to something else - need input:
@scottalanmiller just out of curiosity, could you provide any arguments against using Hyper-V?
We are 99.9% Windows PC & Server shop where I work so naturally some might suggest us using Microsoft's Hyper-V. I have used it a handful of times in the past but it didn't seem very user friendly and seemed to have issues at the time, granted it was over 8 years ago.
Sure...
-
Microsoft no longer has faith in it, it would border on insane to trust as your core platform something that even its own vendor gave up on. The only production style deployment was removed and now only the dev / hobby style is still available. People argue "they didn't discontinue it" and no, not technically, but they did discontinue it from anything that could be argued to be a production methodology. Using HyperV means defining your virtualization infrastructure as a "hobby" environment where uptime and reliability don't matter.
-
It doesn't have the knowledge, development or care of a production system. Never did, but now its a tiny fraction of what it once was. So if you thought it was bad back when MS still pushed it, image how bad it is now!
-
No support. It's an MS product, that means you are left to your own devices and the third party support world is essentially entirely resellers, not skilled engineers. No serious support engineer works with HyperV, so you are depending on the market of somewhat technical salesmen to be your support network.
-
Licensing. Pure and simple, if your infrastructure product requires a license, it's a joke. Nothing causing outages like a license. That's why no serious business will touch that stuff. Risk for no reason is literally a form of sabotage.
-
Confusion. While someone, somewhere installed HyperV and knew what it was, you'll never find that person. Everyone who recommends and implements it does so out of thinking it is something that it is not. Many swear it is a Type2 (and then think that that is a good thing.) Almost no one can actually identify what is virtual and what is not when using it. Especially if you have a Windows show, the chances that anyone really will know HyperV is zero. Only very senior UNIX shops ever have the knowledge to really understand it and they avoid it based on that knowledge.
-
Cost. HyperV isn't free anymore and while someone might argue that you are SO tied to Windows that you'll never be able to make decisions based on what's best for your business, that's a horrific financial mentality to take. Again, risk without benefit, even if small, is very bad. It's what IT pros should be paid to protect against, not recommend. It's a product that only works against your interest, not for them.
-
Management. Like most Windows products, Hyperv's management is a sh1tshow and a completely joke compared to ProxMox and XCP-ng as examples. Not even in the game. If you thought VMware was bad, HyperV is the only thing that gives it serious competition for being so bad.
-
Source and Security. HyperV is closed source and from a notoriously unstable, insecure vendor. You might be stuck with Windows servers, but there's no reason to extend those problems into other arenas without benefit. One mistake does not justify another. Saying "we didn't care about price, security, auditability, performance, stability, etc. with our servers, why should we care now" is the antithesis of our jobs. It's worth noting that all those things clearly aren't top priority, so there might be other factors that we can weigh more heavily. But HyperV lacks ANY benefit (not even familiarity or ease of use!) to Windows admins (ProxMox is literally easier and more straightforward out of the box) so even the seemingly obvious factors work against it. So while giving low weight to production factors is okay, lacking other factors to outweight them doesn't change the outcome.
-
You'd never be able to feel proud of your environment or tell anyone what you run on without feeling embarrassed.
-
Backups, remote support, remote monitoring, etc. all harder. ProxMox and XCP-ng shine at how easy they are to automate, to manually manage, to manage from anywhere. HyperV falls flat in all those areas.
-
THis is a cheat benefit (IT should NOT consider this as it is cheating the company) but HyperV is a valueless skill. You don't want to have that on your resume instead of something of value. So every single person in your IT department should be up in arms at the idea because nothing about it makes their lives easier, and nothing about it is good for their personal careers. Everyone loses, no winners. Again... as IT pros, we should be taking the value to the business first and not making decisions based on what is good for us personally. But this is a care where those two things align.
-
Performance. Haven't checked this recently as HyperV is abandoned in production so it doesn't matter, but historically KVM was the performance leader for Windows workloads.
Most importantly, I can't think of any reason someone would give for even considering HyperV. The utter lack of upsides, being the sole product that EXISTS that's actually worse than, rather than better than, VMware is really something.
Thanks Scott. I was more or less assuming similar to most of the points you listed, but I'm not super knowledgeable or up to date in this area, so this helps me quite a bit. I am leaning towards Proxmox but I still need to show on paper that I've done my due diligence to evaluate each option and all that.
-
-
Another question: when I was researching Proxmox, someone mentioned that it doesn't fully support shared block storage currently. It was basically stated that Proxmox and others haven't come up with an equivalent to VMFS for shared block storage yet, so they are typically leveraging LVM to partition off portions of disk for each VM limit access to those regions to a singular host at a time.
I had looked at this comparison matrix which shows that Proxmox does fully support shared storage, so I'm unclear on the exact specifics and if it really matters in my situation. We basically have an iSCSI storage controller for VM storage and then our ESXi hosts for compute (mentioned in my original post).
All I really care about if we move to Proxmox is that we can store VMs in our storage controller and use the hosts for compute, similar to how we're doing it with VMware today.
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
Another question: when I was researching Proxmox, someone mentioned that it doesn't fully support shared block storage currently. It was basically stated that Proxmox and others haven't come up with an equivalent to VMFS for shared block storage yet, so they are typically leveraging LVM to partition off portions of disk for each VM limit access to those regions to a singular host at a time.
I had looked at this comparison matrix which shows that Proxmox does fully support shared storage, so I'm unclear on the exact specifics and if it really matters in my situation. We basically have an iSCSI storage controller for VM storage and then our ESXi hosts for compute (mentioned in my original post).
All I really care about if we move to Proxmox is that we can store VMs in our storage controller and use the hosts for compute, similar to how we're doing it with VMware today.
The short version is, those people don't know what they're talking about.
Those are two completely different things, with next to no similarities. VMFS is a shared filesystem (better compared to something like Gluster.) LVM is a volume management layer that a filesystem sits on top of.
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
Another question: when I was researching Proxmox, someone mentioned that it doesn't fully support shared block storage currently. It was basically stated that Proxmox and others haven't come up with an equivalent to VMFS for shared block storage yet, so they are typically leveraging LVM to partition off portions of disk for each VM limit access to those regions to a singular host at a time.
That's a bizarre claim, and definitely false. ProxMox supports whatever Linux supports and Linux has had broad support for clustered FS (shared block storage) for at least a decade before VMware even existed.
Plus ProxMox can support VMFS itself! I wouldn't, I'd never use VMware tech BUT, its hilarious because ProxMox (and XCP) have the broadest shared block storage portfolios on the market.
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
I had looked at this comparison matrix which shows that Proxmox does fully support shared storage, so I'm unclear on the exact specifics and if it really matters in my situation. We basically have an iSCSI storage controller for VM storage and then our ESXi hosts for compute (mentioned in my original post).
SOmeone likely doesn't know what shared block is and thinks that VMFS is special or new rather than just VMware's version of something that's been standard a LONG time.
-
@dave247 said in Moving off VMware Hypervisor to something else - need input:
All I really care about if we move to Proxmox is that we can store VMs in our storage controller and use the hosts for compute, similar to how we're doing it with VMware today.
That's actually just a question of iSCSI, NFS or whatever, not even shared block.
-
@travisdh1 said in Moving off VMware Hypervisor to something else - need input:
Those are two completely different things, with next to no similarities. VMFS is a shared filesystem (better compared to something like Gluster.) LVM is a volume management layer that a filesystem sits on top of.
Exactly. Two unrelated topics. I think what people were trying to say is that many people don't do shared storage but do sharded storage instead on ProxMox. And I say "yes, that's true, because most of us on ProxMox always don't want the problems of shared storage."
VMware people do as they are sold. ProxMox people design for what's best. Therefore are less likely to use techniques that aren't broadly useful unnecessarily when alternatives might be simpler.
Really, that person was likely describing the deficiencies in the VMware support ecosystem broadly, but was part of that deficit and didn't realize it.
