KVM vs Hyper-v 2012 Performances

S

starnetwork

Renowned Member
Dec 8, 2009
422
10
83
Hi,
right now am using Openvz (under Proxmox) for Linux VMs and VMware ESXi for Windows VM
and I want to move from VMware to other platform
options:
1. KVM
2. MS Hyper-v 2012

Considering I've already work with Proxmox for Openvz and the Differences in performance between KVM and Hyper-v 2012 (if any - I don't know)
What would you recommend me to work?
KVM or Hyper-v 2012?

Best Regards,
Star Network.
 
I used Hyper-V on 2008 R2 - its a beast for overhead as you have a full fat windows install eating up CPU and Memory before you get to VMs.

I'm very happy with my Proxmox KVM system and wouldnt go back.
 
hotwired007 said:
I used Hyper-V on 2008 R2 - its a beast for overhead as you have a full fat windows install eating up CPU and Memory before you get to VMs.

I'm very happy with my Proxmox KVM system and wouldnt go back.
Click to expand...

Thanks for the Information!
As far as I know Microsoft says the new version is a completely different and better from the previous version, believe them?

Best Regards,
Star Network.
 
You'll still have to have a MS server running at the top layer before the virtual layer - potential security risk and easy access to VM disk storage...

the and AFAIK HyperV is paravirtualisation rather than full virtualisation... and i also seem to remember issues with linux distros being installed...
 
Problems is one thing
but linux is not interested me on this (am working with Proxmox with Openvz for linux and am happy)

Best Regards,
Star Network.
 
starnetwork said:
Problems is one thing
but linux is not interested me on this (am working with Proxmox with Openvz for linux and am happy)

Best Regards,
Star Network.
Click to expand...

I'm running a lot of windows database servers without real good performance.

New version of kvm have some guest hyper-v cpu features for (for windows > 2008), They are not yet added to proxmox,
but I'll try to test them soon.
(This advertise windows vm that they are running in a virtual environnement , and some windows functions works better, like timer management for example).
cpu options are : +hv_relaxed ,+hv_vapic.
 
I don't think Hyper-V can use NFS mounts and requires Microsoft Storage server instead. Consider that in your cost.
 
rugby said:
I don't think Hyper-V can use NFS mounts and requires Microsoft Storage server instead. Consider that in your cost.
Click to expand...

I had to create an account just to correct these posts. If you really wanted to use NFS (which is inferior to SMB3) then Windows Server 2012 is currently the fastest NFS server that there is, and AFAIK the only commercially supported NFS 4.1 implementation - and is fully clusterable, etc. better than any Unix based solution I have ever seen.

- - - Updated - - -

hotwired007 said:
I used Hyper-V on 2008 R2 - its a beast for overhead as you have a full fat windows install eating up CPU and Memory before you get to VMs.

I'm very happy with my Proxmox KVM system and wouldnt go back.
Click to expand...
Then run Hyper-V server which has no Windows OS and is totally free. Hyper-V significantly outperforms KVM in all the recent benchmarks I can find.

- - - Updated - - -

hotwired007 said:
You'll still have to have a MS server running at the top layer before the virtual layer - potential security risk and easy access to VM disk storage...

the and AFAIK HyperV is paravirtualisation rather than full virtualisation... and i also seem to remember issues with linux distros being installed...
Click to expand...
No you don't have to run an MS Windows Server (which incidentally has far fewer security vulnerabilities than any commercial Linux distribution or than VMware). It is full virtualisation. It runs Linux just fine.
 
Hi Richto,
thanks for this grate information, can you please share your benchmarks so we can see the Meaning \ Percent Change ?

Best Regards,
Star Network.
 
Richto,

Thanks for the clarification. I'll download and give this all a run through in our test lab.
 
And still no word what guest OS were used. I bet if the guest OS was Windows that Hyper-V has cheated big time! The disk used for storage SATA can only support this kind of transfer rate if you rewrite the lawes of nature!
 
I don't think that just looking at the performance difference is enough to make a decision.
You have to look at the "big picture" and the overall solution. Meaning you need also to consider things like:

- followup costs (Backup solution etc.)
- management of the system (browser based or client software required)
- how fast can you get to a working state after a crash
- how ist the support (community and/or payed support)
- compatibility with existing hardware

to just name a few.

We are talking about software which is daily changing and improving. Just because of some minor percentage difference in performance today I would not consider changing a working environment.
For me the best solution is pve. I compared vmware and microsoft and pve many times and looking at the "big picture" pve (KVM) was and is the best solution for my needs.

my 2 cents.

best regards
 
We are running Hyper-V 2008 R2 alongside Proxmox 2.3 on similarly configured servers, and Hyper-V performs admirably. Good management, everything is seamless and fast.

Couple of notes about performance:
- Windows guests (with the Hyper-V guest additions installed) are very fast, disk and network IO is comparable to host
- Linux guests (with the Hyper-V kernel additions) are also very fast
- We have even tried to run Proxmox over Hyper-V (without the kernel additions) once as a backup, and when we needed it was able to run one of our heavy MySQL OpenVZ guests without any performance problems (over slow emulated network and IDE disk) for a couple of days

Don't really have a comparison to KVM as we mostly run OpenVZ guests on Proxmox, but I can do a couple of benchmarks if you specify what kind of measurements you are interested in.
 
Last edited:
printsc.png

host: install with default settings, hardware raid 10 4x sas 10k rpm.
guest: 2cores/1socket, 2GB ram, 32GB (pve: virtio lvm, hyper-v: default ide).
no big gap between pve and ms.
 
It would appear that hyper-v is doing write caching in the posted benchmark... While this undoubtedly improves io performance significantly, it carries an extremely high risk of causing data corruption!
 
thats not necessarily true. however: a benchmark that doesnt circumvent caching is completely pointless. you have to always write more test data than the storage target has cache size for a fair comparison.

backtracking to write caching: ceph uses SSD write caches, there no risk involved with that.
 
You must log in or register to reply here.

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!

Quick Navigation

Home Get Subscription Wiki Downloads Proxmox Customer Portal About

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!
Community platform by XenForo® © 2010-2024 XenForo Ltd.
Top Bottom
Back