Survey: Proxmox VE Kernel with or without OpenVZ?
- Thread starter martin
- Start date
We are a 100% KVM shop with over 20 servers running Proxmox. My #1 priority is KVM improvements. OpenVZ is not a priority for us.
With all my respect, I can't believe we are even asking this here. For me PVE relationship with OpenVZ is like a marriage, PVE is nothing without OpenVZ!
Theres bunch of options out there for full virtualization like KVM, even including web interface and open-source options.
But for OpenVZ the options are very limited and poor. PVE is the max authority when you talk about OpenVZ implementation and management.
I would like to vote at least 10 times in this pool to keep OpenVZ.
Regards,
chopeta
Theres bunch of options out there for full virtualization like KVM, even including web interface and open-source options.
But for OpenVZ the options are very limited and poor. PVE is the max authority when you talk about OpenVZ implementation and management.
I would like to vote at least 10 times in this pool to keep OpenVZ.
Regards,
chopeta
I have to vote OpenVZ... In some environments (ie. education or other non-profit) it is not possible to buy the latest servers with VT-x and loads of RAM. In this way Proxmox with OpenVZ lets us really make the most of limited hardware resources. For example we host websites for 50 individual schools. Try running 50 kvm machines on two hosts (dual Xeon, 4GB ram), and compare to 50 openvz containers on the same two hosts.
I found Proxmox when looking for an easy management interface for OpenVZ, it's great to be able to bring up a fully functional server in a few seconds.
On the other hand though, at home I'm trying to get MythTV running on my Proxmox server (not the intended purpose, I know....) and I need kernel 2.6.26 or later for v4l-dvb.
I found Proxmox when looking for an easy management interface for OpenVZ, it's great to be able to bring up a fully functional server in a few seconds.
On the other hand though, at home I'm trying to get MythTV running on my Proxmox server (not the intended purpose, I know....) and I need kernel 2.6.26 or later for v4l-dvb.
I think it would be better to spend time pushing the OpenVZ devs into supporting a newer kernel. They can't avoid it forever. 
I, personally, couldn't live without OpenVZ. Other virtualization solutions are just too slow.
I, personally, couldn't live without OpenVZ. Other virtualization solutions are just too slow.
I voted for using the latest kernel.
We mainly run Win2003/2008 machines on proxmox, thus using kvm. Having OpenVZ running on proxmox means we're bound to an old kernel, crippling kvm (broken virtio because of fairsched patch), no pci device passthrough because of old kernel and all those kind of things.
If it was all up to me I'd drop OpenVZ for now and focus on KVM until OpenVZ catches up to a recent kernel.
We mainly run Win2003/2008 machines on proxmox, thus using kvm. Having OpenVZ running on proxmox means we're bound to an old kernel, crippling kvm (broken virtio because of fairsched patch), no pci device passthrough because of old kernel and all those kind of things.
If it was all up to me I'd drop OpenVZ for now and focus on KVM until OpenVZ catches up to a recent kernel.
Doesn't it seem more logical to switch to a different, KVM-specific, management interface instead of asking the Proxmox team to abandon more than half their user base?
We use PVE for OpenVZ only.
If PVE was just KVM we would have chosen XEN.
You gave up high about the quality of OpenVZ at the time so I'm shocked hearing you even considering dropping it!
We made a choice less then a year ago to go with Proxmox for all our servers because of the professional approach and quality we saw at Proxmox.
Losing it now would be a disaster to us and mean we have to switch to Parallels after all time and efford we invested to avoid this.
Maybe the marriage between KVM and OpenVZ in one interface just isn't practical.
In that case I hope you will develop a separate product for KVM because to us (and most others as I read it here) PVE = OpenVZ.
If PVE was just KVM we would have chosen XEN.
You gave up high about the quality of OpenVZ at the time so I'm shocked hearing you even considering dropping it!
We made a choice less then a year ago to go with Proxmox for all our servers because of the professional approach and quality we saw at Proxmox.
Losing it now would be a disaster to us and mean we have to switch to Parallels after all time and efford we invested to avoid this.
Maybe the marriage between KVM and OpenVZ in one interface just isn't practical.
In that case I hope you will develop a separate product for KVM because to us (and most others as I read it here) PVE = OpenVZ.
Again, this is a survey only. There is no plan to eliminate anything. There is no need to despair.
G
garrettwp
Guest
I voted for latest kernel. I only use KVM. The reason for my choice of the latest kernel is for the latest/updated drivers as this is needed for new and additional hardware that 2.6.24 does not support. The latest kernel will also contain newer filesystems i.e. ext4 and btrfs when it becomes available. The newer kernels will also contain newer versions of KVM as well. In the near future, it will be harder to maintain a good product with an older kernel. I hope the OpenVZ team will come out with support for the newer kernels to prevent Proxmox from falling behind. The main reason for using Proxmox is the web interface. The web interface is key. Otherwise I would have stuck with using virt-manager as the frontend to managing my kvms. I may very well go back to this route if proxmox does not contain a newer kernel release in the near future.
- Garrett
- Garrett
Does it seem logical to stop implementing any core improvements to one of the main virtualization engines in a distribution because the other one fails to keep up or is at least not getting implemented?
If that's the route Proxmox is going, I'd indeed better be looking for a different KVM-only-based distribution with easy install and decent web interface. Any suggestions?
Does it seem logical to stop implementing any core improvements to one of the main virtualization engines in a distribution because the other one fails to keep up or is at least not getting implemented?
If that's the route Proxmox is going, I'd indeed better be looking for a different KVM-only-based distribution with easy install and decent web interface. Any suggestions?
Proxmox VE has good hardware support, a LOT of back ported drivers and very new KVM version - and we will keep container virtualization. there is no plan to remove anything.
I think we need to keep in mind that Openvz and KVM are totally different beasts, they complement rather than compete with each other.
Making use of BOTH of them under one environment is what makes Proxmox unique.
Yes, the old kernel is causing me pain (e.g. I really do need PCI pass-thru), but not as much pain as if Openvz disappeared (e.g. having individual kernels and disk images for multiple VPSs when they are all running the same OS in a low security environment is a massive waste of resources).
Now that this problem with the old kernel has been brought to the surface, I think we should spend some time investigating all possible ways of maintaining this synergy between the 2 virtualization technologies (from running Openvz inside KVM to replacing Openvz with Vserver etc. etc.) before dropping Container based virtualisation from Proxmox altogether.
Making use of BOTH of them under one environment is what makes Proxmox unique.
Yes, the old kernel is causing me pain (e.g. I really do need PCI pass-thru), but not as much pain as if Openvz disappeared (e.g. having individual kernels and disk images for multiple VPSs when they are all running the same OS in a low security environment is a massive waste of resources).
Now that this problem with the old kernel has been brought to the surface, I think we should spend some time investigating all possible ways of maintaining this synergy between the 2 virtualization technologies (from running Openvz inside KVM to replacing Openvz with Vserver etc. etc.) before dropping Container based virtualisation from Proxmox altogether.
Last edited:
I agree with the 90% of people who have stated loosing OpenVZ would be an utter disaster.
I looked to Proxmox VE after looking at VTonf for an alternative to Virtuozzo. Paravirtualization is still _the_ most efficient way to run your (linux based) virtual servers. And an argument that only old servers would use OpenVZ (and thus not green) is nonsense... if you run it on a new powerful server it means you can run even more virtual servers on that 1 server than you can if you used KVM or heaven forbid VMWare ESXi.
But if you find yourself for whatever reason needing a Windows server, Proxmox has you sorted too!
For me KVM is the extra bolt on to Proxmox (it doesn't even need a CPU with VT-x/AMD-V for Proxmox VE to install) that is there as a "just in case".
Proxmox should continue to lead the way in OpenVZ based control panels and work even closer with OpenVZ to ensure it's continuation. Because at the end of the day it's better OpenVZ/Proxmox (both open source) dominate the paravirtualization world than closed source crap like HyperVM that just get hacked.
I looked to Proxmox VE after looking at VTonf for an alternative to Virtuozzo. Paravirtualization is still _the_ most efficient way to run your (linux based) virtual servers. And an argument that only old servers would use OpenVZ (and thus not green) is nonsense... if you run it on a new powerful server it means you can run even more virtual servers on that 1 server than you can if you used KVM or heaven forbid VMWare ESXi.
But if you find yourself for whatever reason needing a Windows server, Proxmox has you sorted too!
For me KVM is the extra bolt on to Proxmox (it doesn't even need a CPU with VT-x/AMD-V for Proxmox VE to install) that is there as a "just in case".
Proxmox should continue to lead the way in OpenVZ based control panels and work even closer with OpenVZ to ensure it's continuation. Because at the end of the day it's better OpenVZ/Proxmox (both open source) dominate the paravirtualization world than closed source crap like HyperVM that just get hacked.
I see from this discussion that a lot of people here are using proxmov-ve with openvz containers. I understand they don't want to lose such an easy and effective tool to manage them as proxmox-ve.
Personally, I am only using KVM, both for Linux and windows machines (but on test for the moment). The most important point for me concerning KVM is that it is developped inside the kernel, so will never be appropriated by one company (even if Red Hat bought Qumranet, other distributions can use it as well). It is this point I think that will make KVM win over Xen or other hypervisor outside the kernel.
The difficulty with other products is to maintain it mainstream with the newest kernels. Red Hat felt very difficult to maintain Xen inside Red Hat, ant Ubuntu too.
It is much more natural to use the last kernels, well integrated, with the latest drivers to install on new machines.
But won't it be possible to have two products : one with a openvz and KVM, but a relatively old kernel, and another one with the latest stock kernel, and without openvz ?
Personally, I am only using KVM, both for Linux and windows machines (but on test for the moment). The most important point for me concerning KVM is that it is developped inside the kernel, so will never be appropriated by one company (even if Red Hat bought Qumranet, other distributions can use it as well). It is this point I think that will make KVM win over Xen or other hypervisor outside the kernel.
The difficulty with other products is to maintain it mainstream with the newest kernels. Red Hat felt very difficult to maintain Xen inside Red Hat, ant Ubuntu too.
It is much more natural to use the last kernels, well integrated, with the latest drivers to install on new machines.
But won't it be possible to have two products : one with a openvz and KVM, but a relatively old kernel, and another one with the latest stock kernel, and without openvz ?
The simple fact that this discussion exists has us looking at the many alternatives on the market- both opensource and pay-to-play.
We like OpenVZ - and we also like KVM.
Xen- i am not convinced is dead -but the writing is on the wall some would say...
For ProxMox to drop OpenVZ would simply - in my opinion- remove the reason for us to continue using ProxMox - (man thats painful to say).
I suspect for many others - that holds true as well.
If we need to change the ropes now - especially with how openVZ is - because of the shared nature of the filesystem and kernel - it is not a trivial matter.
Because of this - we have already begun the process of moving off of OpenVZ ... all new clients have other options including vmware -
Ashame.
We like OpenVZ - and we also like KVM.
Xen- i am not convinced is dead -but the writing is on the wall some would say...
For ProxMox to drop OpenVZ would simply - in my opinion- remove the reason for us to continue using ProxMox - (man thats painful to say).
I suspect for many others - that holds true as well.
If we need to change the ropes now - especially with how openVZ is - because of the shared nature of the filesystem and kernel - it is not a trivial matter.
Because of this - we have already begun the process of moving off of OpenVZ ... all new clients have other options including vmware -
Ashame.
Amen. We are running OpenVZ on several very modern, very updated servers. I would have to cast my vote for keeping OpenVZ, KVM is nice, but we use that aspect of the OS minimally. OpenVZ allows us to maximize speed and efficiency while being minimal on hardware resources.
If ProxMox were a woman, I would marry it. Thanks fellas!
Amen. We are running OpenVZ on several very modern, very updated servers. I would have to cast my vote for keeping OpenVZ, KVM is nice, but we use that aspect of the OS minimally. OpenVZ allows us to maximize speed and efficiency while being minimal on hardware resources.
If ProxMox were a woman, I would marry it. Thanks fellas!
Even on my servers i use OPENVZ and i need it.
For the marriage just a question:
would you marry ProxMox due to the silence of the hardware?
Diaolin
lol, a true professional never tellsEven on my servers i use OPENVZ and i need it.
For the marriage just a question:
would you marry ProxMox due to the silence of the hardware?
Diaolin