Forum for linux virtualization?

[s0lid]

We already got forums for commercial virtualization solutions such as VMware/Virtualbox/Citrix and Windows Server/Hyper-V.
There should be 3rd forum for Linux kernel enabled virtualization such as KVM/XEN.

 

[Stanza]

here here, yes please

 

[Chuckleb]

I agree as well!

 

[Patrick]

What about just merging to a virtualization forum?

 

[Diavuno]

Nah, I like categorized

 

[s0lid]

Categorized would be better. I just think that the kvm/xen threads would be buried under hyper-v/esxi threads.

 

[Diavuno]

Categorized would be better. I just think that the kvm/xen threads would be buried under hyper-v/esxi threads.

probably...

 

[dealcorn]

I wish it were so, but I do not think community support is strong enough to let a KVM forum succeed. I recall a thread a year or two ago asking if anyone used KVM. It was never answered. I plan to use KVM to dynamically load a windows VM acting as a DVR, a PCI-e tuner is in transit. Linux currently does not support the analog side of any PCI-e TV tuner. Many/most Linux users need a Windows license to run at least one Windows application and a VM solution is less troublesome than dual booting Windows with Grub. However, this has nothing to do with servers.

Does any SOHO market for KVM exist? Several forum members have constructed sophisticated hypervisor environments that use KVM competitors. Their thoughts on why KVM was not selected might be helpful. I suspect that selection of a "known good" environment with familiar/superior support tools matters more than the generally immaterial performance benefits of KVM. The server room is conservative for good reasons. I expect that broader KVM deployment by the big boys will precede market acceptance in the SOHO market.

On the other hand, the economic benefits of containers are far easier to understand and broad market acceptance is a slam dunk as the technology gets better understood. A Docker article may be on Patrick's to do list. I use Shorewall as a firewall with NAT and the only reason it is not in a container is personal ignorance. From a security perspective, I suspect there is a real benefit from implementing my P2P daemon in it's own container also. If KVM winds up in a separate forum, I would also include containers as part of the Linux alternatives to commercial hypervisors.

 

[nry]

Personally I plan on playing with KVM a little with OpenStack but that's quite far down the todo list at the moment!

 

[s0lid]

I'm planning to run tad modified Proxmox after I merge my solaris and ESXi servers into one machine.
We had good test run at this years Vectorama where we were running ZFS natively on proxmox using self configured 3.15 kernel on a Dell VRTX node.

Proxmox is Debian based KVM-hypervisor and you can use debian repos on it.
So my plan for the mordor(hostname) is:
- Install ZoL from sources
- Compile newest kernel since I need the Brocade 1020 Drivers
- Fill the server to the brim, add second L5630, Add 48GB of ram to it and fill the 2 empty hdds bays with 3TB DT01ACAs.
- Convert the current 8x3TB Raidz2 into 10x3TB mirror-stripe(raid10) pool.

 

[MiniKnight]

I'd say 1 forum.

 

[Chuckleb]

I've moving heavily into Docker for projects at work so have an interest in containerization. We are using it as a means to deploy complex applications.

 

[Mike]

I wish it were so, but I do not think community support is strong enough to let a KVM forum succeed. I recall a thread a year or two ago asking if anyone used KVM. It was never answered. I plan to use KVM to dynamically load a windows VM acting as a DVR, a PCI-e tuner is in transit. Linux currently does not support the analog side of any PCI-e TV tuner. Many/most Linux users need a Windows license to run at least one Windows application and a VM solution is less troublesome than dual booting Windows with Grub. However, this has nothing to do with servers.

Does any SOHO market for KVM exist? Several forum members have constructed sophisticated hypervisor environments that use KVM competitors. Their thoughts on why KVM was not selected might be helpful. I suspect that selection of a "known good" environment with familiar/superior support tools matters more than the generally immaterial performance benefits of KVM. The server room is conservative for good reasons. I expect that broader KVM deployment by the big boys will precede market acceptance in the SOHO market.

On the other hand, the economic benefits of containers are far easier to understand and broad market acceptance is a slam dunk as the technology gets better understood. A Docker article may be on Patrick's to do list. I use Shorewall as a firewall with NAT and the only reason it is not in a container is personal ignorance. From a security perspective, I suspect there is a real benefit from implementing my P2P daemon in it's own container also. If KVM winds up in a separate forum, I would also include containers as part of the Linux alternatives to commercial hypervisors.

On the contrary, i find it extremely weird that everybody sticks with the Vmwarez products for home use. People are making virtual machines for storage, which boot from the locked down vmwarez FS, then export that over less-than-optimal transport protocols like NFS back to the host to have other machines boot from the COW FS, albeit without direct access to the COW capabilities you would have natively. The kernels these products run have poor hardware support and are all but efficient. The software that runs on this host kernel is very hard to extend, outdated and all but updatable by the user.
So, the upside to this may be that you gain some experience with this point-and-click enterprise product.

 

[RTM]

What about just merging to a virtualization forum?

I kind of like this idea, its not like there is that many threads in the respective forums anyway.
And at the moment it is kind of difficult to place generic threads, like "what virtualization platform to use".

Perhaps the solution is to make subforums, like VMware, Linux, Microsoft, etc under a virtualization forum.
Same could be done for NAS software, and maybe for generic OS.

 

[Diavuno]

On the contrary, i find it extremely weird that everybody sticks with the Vmwarez products for home use. People are making virtual machines for storage, which boot from the locked down vmwarez FS, then export that over less-than-optimal transport protocols like NFS back to the host to have other machines boot from the COW FS, albeit without direct access to the COW capabilities you would have natively. The kernels these products run have poor hardware support and are all but efficient. The software that runs on this host kernel is very hard to extend, outdated and all but updatable by the user.
So, the upside to this may be that you gain some experience with this point-and-click enterprise product.

there is a reason most stick to vmware...

they might not be as open as kvm/others but: THEY JUST WORK.

I remember playing with esx years ago on a Dell 2600? it was a dual 604 tower that was before VT, or VT-IO even before x64 I think. (not sure about the last one!)

vmware could do things better, sure.
but any VM works, and pretty fast considering the tech behind it!
what they do they do well.

free is nice for home users too.

also keep in mind anyone with a VM host isnt a consumer, probably not even a "prosumer" but moth than likely a computer something...
most people are bias to what they are comfortable with and VM ware still dominates enterprise and SMB

 

[Cheddoleum]

+1 on this. The whole KVM / QEMU / Xen / LXC ecosystem is extremely hot right now, the standard of work is exceptionally high for bleeding edge FOSS, with a lot of very sane and sensible convergence on common tools and libraries.

Lately I've been migrating a lot of stuff to LXC that I used to use KVM for; it's particularly useful when you want to share some amount of common storage without introducing a network filesystem layer, plus with a common kernel you don't need hardware support for "stealing" devices, and you don't have to provision memory. The main contraindication in my book is if the application requires me to introduce a lot of funky modules into the host kernel that it wouldn't otherwise need.