Survey: Proxmox VE Kernel with or without OpenVZ?
- Thread starter martin
- Start date
I've been going through the hard path of evaluation several solutions, including VMWare (the free Server and also ESXi 4), Ubuntu Enterprise Cloud, Eucalyptus on a plain Debian machine, conVirt and also looked at oVirt.
The biggest advantage of Proxmox VE for me is the support for OpenVZ and KVM (which none of the aforementioned solutions do) at the same time. The easy to use GUI is also a plus, although other solutions offer a similar experience (though not for managing OpenVZ, but rather KVM and XEN).
I mostly use OpenVZ with Linux appliances and only a few KVMs with Windows (sigh) and I am more interested in a reliable, scalable and stable Linux virtualization environment where full virt isn't a must.
Dropping OpenVZ from PVE would mean I'd have to find something else :-(
The biggest advantage of Proxmox VE for me is the support for OpenVZ and KVM (which none of the aforementioned solutions do) at the same time. The easy to use GUI is also a plus, although other solutions offer a similar experience (though not for managing OpenVZ, but rather KVM and XEN).
I mostly use OpenVZ with Linux appliances and only a few KVMs with Windows (sigh) and I am more interested in a reliable, scalable and stable Linux virtualization environment where full virt isn't a must.
Dropping OpenVZ from PVE would mean I'd have to find something else :-(
Has any one looked at "Linux Containers" as a replacement for OpenVZ. its a part of the lastest kernels and seems to do most of the things OpenVZ can do.
http://lxc.sourceforge.net/
Maybe the better option might be to build capability for this into Proxmoz and over the phase out openVZ. Would mean having two kernels for awhile.
Mike
http://lxc.sourceforge.net/
Maybe the better option might be to build capability for this into Proxmoz and over the phase out openVZ. Would mean having two kernels for awhile.
Mike
I'd certainly would miss OpenVZ. I've got quite a few machines running on it. It's super fast to set up, and simple to manage. I love it.
That said, should it be determined that it needs to go, please at least provide some "streamlined" way of migrating OpenVZ to KVM machines. Reading through this thread, a lot of people, just like me, have a lot of OpenVZ machines running on PVE. I don't even want to think about the time sink it would mean, migrating all the OpenVZ machines to KVM.
I must say, if needed and possible, I like the idea of two versions. Install OpenVZ or KVM. Then, keep the management interface, allowing admins to manage all their PVE hosts in a cluster, OpenVZ or KVM, from the master machine.
That said, should it be determined that it needs to go, please at least provide some "streamlined" way of migrating OpenVZ to KVM machines. Reading through this thread, a lot of people, just like me, have a lot of OpenVZ machines running on PVE. I don't even want to think about the time sink it would mean, migrating all the OpenVZ machines to KVM.
I must say, if needed and possible, I like the idea of two versions. Install OpenVZ or KVM. Then, keep the management interface, allowing admins to manage all their PVE hosts in a cluster, OpenVZ or KVM, from the master machine.
This is a really good idea, keep the option for OpenVZ by simple having separate machines which support different system but are managed by the same GUI.
Mike
It may not be, I've given it a try and it seems to work ok, but that's not really the point. Your looking for a solution in the medium to long term.
For the same reason you support KVM not XEN, we need to look at the preferred option which is LXC.
hope I've not being to forcefulThanks for your product its great
Mike
I fail to see why we cant have the choice to run two different kernels.
eg. if i choose to run the latest vanilla kernel that just has kvm support then the gui should have all openvz related config disabled.
Im quite happy to put my hand up to spend some development time to get this working in this manner
eg. if i choose to run the latest vanilla kernel that just has kvm support then the gui should have all openvz related config disabled.
Im quite happy to put my hand up to spend some development time to get this working in this manner
For the same reason you support KVM not XEN, we need to look at the preferred option which is LXC.
You missed the fact that we already work on LXC and provided patches to the project ;-)
Searching on NET LXC versus OpenVZ.
OpenVZ is more mature then LXC and have more resources.
my sugestion is whait linux contaniers keep more stable and usefull.
Debian minimal cycle is 2 years, why not whait and keep old kernel?
Keep Stability and give more time to work on a solution provide 2 solutions?
I see ubuntu will be suport LXC + KVM and a manager interface. i don´t no more about this. but i belive proxmox is a great solution using 2 tecnologies.
On a ideal world, I can see proxmox official debian virtualization solution with a instalation profile and tasksel to choose it on setup. it´s just a dream.
OpenVZ is more mature then LXC and have more resources.
my sugestion is whait linux contaniers keep more stable and usefull.
Debian minimal cycle is 2 years, why not whait and keep old kernel?
Keep Stability and give more time to work on a solution provide 2 solutions?
I see ubuntu will be suport LXC + KVM and a manager interface. i don´t no more about this. but i belive proxmox is a great solution using 2 tecnologies.
On a ideal world, I can see proxmox official debian virtualization solution with a instalation profile and tasksel to choose it on setup. it´s just a dream.
N
Ninjix
Guest
I've been evaluating Proxmox for the last week on my test rig with an older RAID controller and I was really impressed with the clean, no-nonsense UI. Everything was going great and I came in this weekend to prepare a Monday demo for my CTO on one of our new production machines which have LSI MegaRAID 9260 cards. Ouch! The old Kernel bit me.
This is really too bad. Proxmox was so close to being the out-of-box bare metal winner. Don't let this wonderful product fall behind. OpenVZ is nice but Proxmox running on a 2.6.32 Kernel with KSM would be nicer. IMHO
Having a Debian based Linux-KVM solution with this level of professional polish that you can buy support is really awesome. Is there anyway Proxmox can be ported to the future Ubuntu 10.04? Then the dev team could concentrate on the software and sell support for it. Canonical can handle support for the OS level. Then two Linux companies can collect my support service contracts.
This is really too bad. Proxmox was so close to being the out-of-box bare metal winner. Don't let this wonderful product fall behind. OpenVZ is nice but Proxmox running on a 2.6.32 Kernel with KSM would be nicer. IMHO
Having a Debian based Linux-KVM solution with this level of professional polish that you can buy support is really awesome. Is there anyway Proxmox can be ported to the future Ubuntu 10.04? Then the dev team could concentrate on the software and sell support for it. Canonical can handle support for the OS level. Then two Linux companies can collect my support service contracts.
N
Ninjix
Guest
::Update::
I couldn't walk away from Proxmox after using it. It is possible to install it on Ubuntu 9.10 Server with a stock kernel.
# uname -a
Linux node-b 2.6.31-16-server #53-Ubuntu SMP Tue Dec 8 05:08:02 UTC 2009 x86_64 GNU/Linux
I use the "Unsupported Debian Lenny" wiki entry as a starter and use the pvetest debs. Most of all the needed packages installed with an apt-get install pve-manager.
Something didn't like the Fair Scheduler "-id xxx -cpuunit 1000" appending to the KVM command so commented it out in Perl script. That's ok because I'm using the VirtIO drivers.
Guys, I know this is way outside of the Proxmox support realm and I'm sure to run into other bugs but wow it easy to get it running on Ubuntu. I am again impressed with Proxmox and the team who develops it. The craftmanship of the Debian packages and the well organized Perl code made this an easy 2 hour exercise.
Now I've got Proxmox PVE running on the latest hardware to show my execs tomorrow.
I want Proxmox PVE-KVM on the 2.6.32.x kernel even more.
I couldn't walk away from Proxmox after using it. It is possible to install it on Ubuntu 9.10 Server with a stock kernel.
# uname -a
Linux node-b 2.6.31-16-server #53-Ubuntu SMP Tue Dec 8 05:08:02 UTC 2009 x86_64 GNU/Linux
I use the "Unsupported Debian Lenny" wiki entry as a starter and use the pvetest debs. Most of all the needed packages installed with an apt-get install pve-manager.
Something didn't like the Fair Scheduler "-id xxx -cpuunit 1000" appending to the KVM command so commented it out in Perl script. That's ok because I'm using the VirtIO drivers.
Guys, I know this is way outside of the Proxmox support realm and I'm sure to run into other bugs but wow it easy to get it running on Ubuntu. I am again impressed with Proxmox and the team who develops it. The craftmanship of the Debian packages and the well organized Perl code made this an easy 2 hour exercise.
Now I've got Proxmox PVE running on the latest hardware to show my execs tomorrow.
I want Proxmox PVE-KVM on the 2.6.32.x kernel even more.
Last edited by a moderator:
We will release a 2.6.32 based kernel for testing soon.
Last edited by a moderator:
N
Ninjix
Guest
Excellent news! Thanks for the reply, dietmar.
Do you have a rough roadmap for Proxmox supported kernel version upgrades? First or second half of 2010?
Do you have a rough roadmap for Proxmox supported kernel version upgrades? First or second half of 2010?
we are already in heavy testing, if all goes well we will release it to public (pvetest repo) this year.