Survey: Proxmox VE Kernel with or without OpenVZ?

Proxmox VE Kernel with or without OpenVZ?

  • Keep old Kernel with OpenVZ support (2.6.24)

    Votes: 143 60.3%
  • Use the latest Linux Kernel (without OpenVZ but with best KVM and hardware support)

    Votes: 94 39.7%

  • Total voters
    237
We use both features, but I would keep 2 separate flavors of proxmox if that's the only way of having the most up to date of both worlds.

In any case, we need the most recent version of kvm possible, we'd better have that tan openvpz. You can always have proxmox on a KVM VM inside proxmox.

My vote is, keep up with KVM if I have to choose. The good thing of proxmox in not only it has OpenVZ AND KVM. KVZ is the only solution (OSS or comercial) that supports 16 vcpus, nor xen nor vmware do that on a windows guest, and we need that feature plus the option to assign cores to guest as logical cores, not sockets. That's what we are waiting to see in KVM in proxmox soon.
 
Full virtualization is the future (and present) so my vote is for KVM only. The older kernels are holding back the feature set available (unless you are back porting, which isn't fun either). While Red Hat has a nice offering, Proxmox is unique in it's all includive install, and easy web interface. I think pushing into a KVM only platform will certainly expand the use of Proxmox.

However, I understand why those using OpenVZ won't want to see it dropped, so perhaps the best solution would be to split the system into two versions - KVM only, and OpenVZ only. Over time though, I would see the OpenVZ version completely replaced by KVM. The development has already slowed, and with so much attention focused on KVM, that will continue to be where rapid progress takes place (as it is now).

BTW - Congrats on the 1.4 Beta - looks great!
 
Full virtualization is the future (and present)

With all respect, but where do you get that idea??

In my opinion, for Linux virtualization I can't see any advantages in full virtualization, on the contrary.
 
Re: Proxmox VE Kernel with or without OpenVZ?

We have run our PBX in a Flash training systems on Proxmox as KVMs for more than a year. We finally got around to releasing an OpenVZ image last week. All I can say is WOW! What a difference in performance and ease of setup, not to mention the huge leap in simultaneous images which can be run on a single server. There may be a place for KVMs with newer kernels, but we'd be very disappointed to see the OpenVZ option disappear. What we really need are more OpenVZ applications, not less. As for the kernel holding folks back, our new OpenVZ template is functionally identical to our very latest CentOS 5.3 build with zero issues and all the RedHat bells and whistles. But don't take my word for it. Download it and take it for a spin yourself.
 
Last edited:
bitbud said:
Full virtualization is the future (and present)
Hm, this depends on your use case. Some prefer full virtualisation (e.g. hosting multiple Windows VM on a Proxmox server) while others run tons of Linux "root servers" on their server. Plus the fact that not all "metal" out there supports KVM.

I think for most plain Linux server stuff OpenVZ is the way to go with the smallest overhead.
Running a Linux VM as KVM instead of OpenVZ container simply means that you can run less servers on your physical host.

dietmar said:
At least that is what some large companies wants you to believe ;-)
Esp. those who do not support container virtualisation :rolleyes:

I think that KVM (be it with Proxmox or not) will be a big player in the near future rivaling with the other "free" VM products out there.
 
We use both solutions, but I guess the real question here is, what do we have to gain by changing kernel. What are the pros and cons of loosing OpenVZ to be able to upgrade kernel and KVM ? Is there a big advantage ?
Why isn't OpenVZ supporting new kernels ? Wont the lack of evolutiona/support of OpenVZ be a problem in the near future ?
 
hi

With my current provider, who blocks traffic that comes from internal MAC addresses it does not know, I'd be stuffed without OpenVZ, or at least I'd be having an administrative nightmare, and support issues trying to explain to my hosting provider what I was trying to do, and why. Definitely keep it.

Joe
 
hi

With my current provider, who blocks traffic that comes from internal MAC addresses it does not know, I'd be stuffed without OpenVZ, or at least I'd be having an administrative nightmare, and support issues trying to explain to my hosting provider what I was trying to do, and why. Definitely keep it.

Joe

for the provider issue:
did you get this?
 
Hi

Thanks for the pointer, I posted this some months ago that you kindly helped me with, but I could not get it to work, however, in my application (Asterisk, VoIP, FreePBX, A2Billing), that forced me to do this which in turn spawned this article, out this morning.

http://nerdvittles.com/

Additionally, I have a work in progress but be gentle, it's my first shell script for Debian, although I am now beginning to see the attractions of Debian over RHEL based releases.



Joe
 
comparatively new to the virtualization game. but really not interested in openvz. the best KVM development/support would be a priority for us...
 
i tried some of the virtualization like vmware, xen, virtualbox, vserver, kvm, openvz and i evaluated openvz and kvm for me;

actually openvz and kvm are running on debian, managed via console and i searched around for gui's to simplify management;

there are some tools out there for openvz (webvz, easyvz) and for kvm (qemuctl, qemu-launcher, qemuctl, qemulator, qtemu, ovirt) - some are good, some are crappy;

nearly three weeks ago i tested proxmox which manages both of my favourite virtualization software and i am really impressed;

web based and os-independent management - great!

proxmox is based on debian, and some of us won't to use something else than debian for some reasons;
debian based systems are known for performance, stability, security, ... but not for the latest features;

if you need the latest features, i recomment to use debian sid, some other distribution with recent updates, build the sources by yourself or buy a software which does all you want....

for those who only want kvm in it's latest version i recommend to build the latest kernel and kvm sources by yourself and use ovirt for management on the distro you prefer;

for those who only want openvz, proxmox is the best and easy to use i tried in performance, features and management;

there is no need to kick one of them, when they are stable enough for production use, it should be integrated in proxmox;

if there is no recent kernel availiable, we have to wait, use the existing version and upgrade if it is availiable;

maybe in the meantime there could some other features be added on the network side for bonding, vlan, trunking, vde2, network utilization, beancounter settings, ....
a role based gui for managing the virtual system would also very cool (the user which owns his vm could start/stop, vnc to his machine, see soem statistics, ....)

go ahaed guys from proxmox, you did a really good job to provide us an opensource virtualization solution with clustering in one web based management gui;

i don't want to miss one of them....
 
Last edited:
The idea of running a KVM host with Proxmox in OpenVZ mode within is interesting... BUT... many of us are running Proxmox on hardware without VT support which means KVM will not work. OpenVZ must be kept... even if that means staying back a bit. Branching into two versions, one for updated KVM and another with OpenVZ would be fine with me...
 
My vote goes for keeping the openvz kernel. It lets me do more for less. I implement solutions for SMB, and openvz is one of my secrets to remain competitive. However, I value being able to use KVM if needed.
¿What do I imagine for the future?

Having

  • A proxomox ve administration console, decoupled from the servers
  • A proxmox ve open-vz server
  • A proxmox ve kvm server
And being able to install any of them, on debian or ubuntu distributions, via apt-get. That will let me have a farm of "up to date" kvm and open-vz servers, and the administration console in any of them, or in an independent server. Yes, may be a lot much more work, but Hey! may be the community can help.
 
¿What do I imagine for the future?

Having

  • A proxomox ve administration console, decoupled from the servers
  • A proxmox ve open-vz server
  • A proxmox ve kvm server

Proxmox VE it's a software based on GPL license...so anyone who want to modify the kernel it's free to do it...and post the results!

To separate PVE in 2 distributions it's the bad ideea I ever ear on this forum!

Maybe a sollution will be a menu where to select the kernel who intend to use!
 
To separate PVE in 2 distributions it's the bad ideea I ever ear on this forum!
Could you elaborate why ? The burden would be mainly for the proxmox team with two branches to maintain and support. But in my opinion, it would be much easier to maintain a branch with only KVM. You don't have to backport drivers, KVM patches from new kernels, and certainly other things.

Maybe a sollution will be a menu where to select the kernel who intend to use!
I think it is a bit more complicated than that. The options in the web interface, and the behavior, would not be the same. At least, there will be no openvz option in the KVM branch !

Alain
 
Imagine that Small companies (with less then 20 employers) don't use powerfull hardware...so for them actual PVE it's perfect!
They can install a Windows SBS Server (KVM), PMG (openVZ), and maybe an linux...so in this way they use only one hardware...one machine!

In your case...they need at least 2 servers!

Maybe for enterprise will be a sollution...but I'm not sure. Try to imagine different situations!
 
Imagine that Small companies (with less then 20 employers) don't use powerfull hardware...so for them actual PVE it's perfect!
They can install a Windows SBS Server (KVM), PMG (openVZ), and maybe an linux...so in this way they use only one hardware...one machine!

In your case...they need at least 2 servers!

No, if the branch with openvz has also a KVM kernel, this is the case with 2.6.24. So you would have an older KVM, but still a KVM option to install a windows server.

And you know, in the near future, all machines, even small laptops, will have CPU with virtualization options available (this is already the case for AMD, and will be mandatory for Intel to support XP mode in Windows 7). So, when older machines will disappear, all machines, even cheap ones, will be able to support KVM...
 
And you know, in the near future, all machines, even small laptops, will have CPU with virtualization options available (this is already the case for AMD, and will be mandatory for Intel to support XP mode in Windows 7). So, when older machines will disappear, all machines, even cheap ones, will be able to support KVM...

That may very well be the case, but being able to SUPPORT kvm, and actually RUNNING kvm are two totally different things.

And although the demise of some older machines is in sight, they are not going to all suddenly disappear mid next year. They will be around for a bit still to come. (IE6, anyone?)

Just because it isn't the best, doesn't mean it is worthless.
 

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!