[SOLVED] Is Proxmox future-proof?

[Razva]

Hey,

I'm studying the market right now in order to find solutions for a semi-public cloud. Here's the result of my small research:
- OpenVZ is a very good solution for most Linux hosts because the vast majority of Linux users don't really need custom kernels and "nuclear rockets". Users are mostly looking for a fast way to build/reinstall/test and - when necessary - expand. Thinking at the fact the iolimit was introduced in recent vzkernel...a lot of concerns are covered;
- KVM is the "new kid on the block" when it comes to Windows virtualization. I'm personally using XenServer, but a RedHat Powered solution sounds very interesting.

So in my mind the best solution is to offer OpenVZ + KVM. Exactly what Proxmox is doing!

But here comes the tricky question: is Proxmox safe?

I don't want to be disrespectful, but it looks like a "one-man show". I'm positive that the creator/maintainer has impressive technical skills and it has my greatest respect! But when it comes to implementing Proxmox company-wide, even as a startup, you have to be very careful what you choose on the long-run. Accidents are common nowadays (just think at yesterday's Paul Walker) so implementing a solution developed by a single person sounds like a high-risk deal.

Just think about HyperVM. They were once the best solution for controling OpenVZ containers, but now the project is mostly dead.

What's your opinion about this? Is Proxmox "safe"? Why?

 

[udo]

Hey,

I'm studying the market right now in order to find solutions for a semi-public cloud. Here's the result of my small research:
- OpenVZ is a very good solution for most Linux hosts because the vast majority of Linux users don't really need custom kernels and "nuclear rockets". Users are mostly looking for a fast way to build/reinstall/test and - when necessary - expand. Thinking at the fact the iolimit was introduced in recent vzkernel...a lot of concerns are covered;
- KVM is the "new kid on the block" when it comes to Windows virtualization. I'm personally using XenServer, but a RedHat Powered solution sounds very interesting.

So in my mind the best solution is to offer OpenVZ + KVM. Exactly what Proxmox is doing!

But here comes the tricky question: is Proxmox safe?

I don't want to be disrespectful, but it looks like a "one-man show". I'm positive that the creator/maintainer has impressive technical skills and it has my greatest respect! But when it comes to implementing Proxmox company-wide, even as a startup, you have to be very careful what you choose on the long-run. Accidents are common nowadays (just think at yesterday's Paul Walker) so implementing a solution developed by a single person sounds like a high-risk deal.

Just think about HyperVM. They were once the best solution for controling OpenVZ containers, but now the project is mostly dead.

What's your opinion about this? Is Proxmox "safe"? Why?

Hi,
proxmox-ve isn't an one man show - it's the poduct of an company which stay since some years on the market.

I use pve since almost 5 years in production and it's stable enough.
And kvm is also not the new kid - this is also some years ago...

Udo

 

[Razva]

My question is very simple: what happens if the head-developer will not have the option to continue working on this project? Can it have continuity?

 

[m.ardito]

Can it have continuity?

having the source available, and relying completely on (proven) open source technologies, you have the maximum chance about its continuity.

But any project, small or big, commercial or open source, can fail in the long term.

Marco

 

[felipe]

i was thinking the same. what i am a little bit afraid is that i think the manpower is not enough for maintaining kvm and openvz at the same time. it seems to me that openvz holds the development of kvm. the proxmox kernel is 2.3.23 at the moment and does not support a lot of new features. for example windows 2012 TS (rds) will not work because it needs nested virtualization. i think that windows environment is not hat important to the proxmox team because windows machins make lot of troubles.... (they still run perfect if you exactly know what you are doing) but in newer kernels they seemed to work with less problems...(more stable network drivers etc..)
and what i could not find is a roadmap for the next versions of proxmox!

 

[adamb]

It doesn't bother me much as I know I could put something together which is very similar with CentOS or just about any distro. Proxmox is great though because it saves me from having to do this.

 

[tom]

i was thinking the same. what i am a little bit afraid is that i think the manpower is not enough for maintaining kvm and openvz at the same time. it seems to me that openvz holds the development of kvm. the proxmox kernel is 2.3.23 at the moment and does not support a lot of new features. for example windows 2012 TS (rds) will not work because it needs nested virtualization. i think that windows environment is not hat important to the proxmox team because windows machins make lot of troubles.... (they still run perfect if you exactly know what you are doing) but in newer kernels they seemed to work with less problems...(more stable network drivers etc..)
and what i could not find is a roadmap for the next versions of proxmox!

See the roadmap, also shows the history of already more than 5 years.
http://pve.proxmox.com/wiki/Roadmap

I appreciate your post but "believing" that we cannot do the job seems just your personal opinion.

Our QEMU/KVM is latest and greatest, 1.7 is already working great (see git version). So most of your statements are just not true.

Our 2.6.32 is based on RHEL6.x, the most used kernel version in enterprise environments worldwide. As soon as RHEL7 beta is here, a new Proxmox VE kernel is also expected.

And yes, nesting does not work with 2.6.32, but will work with the RHEL7 kernel.

 

[felipe]

i am not believing that you can not do your job!
You are doing a great job. i just see that its a loooot of work to do all this stuff (specially that openvz & kvm works with mostly all features) and the proxomox team is not that big.
i was just wondering for example that the rtl8139 card had a lot more problems in proxmox then before. ok its not a good idea to use it... but as i never had problems before i never changed them on old smaller machines... had another strange behavior which just happend with the proxmox kernel... but i could solve it.
i really like your efforts for all the cluster and backup stuff i just love it.

 

[tom]

i am not believing that you can not do your job!
You are doing a great job. i just see that its a loooot of work to do all this stuff (specially that openvz & kvm works with mostly all features) and the proxomox team is not that big.
i was just wondering for example that the rtl8139 card had a lot more problems in proxmox then before. ok its not a good idea to use it... but as i never had problems before i never changed them on old smaller machines... had another strange behavior which just happend with the proxmox kernel... but i could solve it.
i really like your efforts for all the cluster and backup stuff i just love it.

If you have issue, there are a lot of places to get help. We even offer help for free, but also commercial support.

So if you have issues, address them instead of "believing" something.

1. Open a new thread in the forum
2. Open a bug report if you find a bug
3. Or go for commercial support

And just a side note: I have daily issue with Adobe and Oracle software, seems these companies seems also too small - or is it just that software is complex and bugs exists everywhere?

 

[Razva]

Hey Tom,

Bugs - like in the human body - exist will continue to come up. That is the normal way of "being".

The difference between Adobe/Oracle and our thread is that I can be 1000% sure that Adobe will provide updates for at least 10-15 years. I can invest resources in training employees, infrastructure and other dependent software, even if their CEO/CTO "dies in a tragic accident" (God forbid!).

So the thread is not about bugs, is about implementing a product company-wide. It's more about long-term "trust" and "relations". Again don't get me wrong, I have big respect for you guys, but I personally don't know much about Proxmox, its employees (if any) and its leadership. I know I can trust Parallels, I know I can trust vmWare, I know I can trust Microsoft and Citrix. So when you - Proxmox - are playing "with the big guys" your users need to know if they can depend on you on the long run.

 

[tom]

Hey Tom,

Bugs - like in the human body - exist will continue to come up. That is the normal way of "being".

yes.

The difference between Adobe/Oracle and our thread is that I can be 1000% sure that Adobe will provide updates for at least 10-15 years.

I do not comment this, but its a good joke. Seems you live in a different world than me.

I can invest resources in training employees, infrastructure and other dependent software, even if their CEO/CTO "dies in a tragic accident" (God forbid!).

So the thread is not about bugs, is about implementing a product company-wide. It's more about long-term "trust" and "relations". Again don't get me wrong, I have big respect for you guys, but I personally don't know much about Proxmox, its employees (if any) and its leadership. I know I can trust Parallels, I know I can trust vmWare, I know I can trust Microsoft and Citrix. So when you - Proxmox - are playing "with the big guys" your users need to know if they can depend on you on the long run.

You trust closed source software from US located companies? You know the patriot act? You know the NSA?
We have a different understanding of trust and relation here, but you are free to trust all these guys. The only guarantee of backdoor free software is open source software.

We were one of the first companies (or the first one?) offering commercial support for KVM based virtualization solutions. We have a long history (see http://pve.proxmox.com/wiki/Roadmap), all our code is open source, we have a quite powerful community.

But at the end, you have to decide by your own, I just thought to add some hard facts and arguments in this thread.

 

[davlaw]

Hey Tom,

So the thread is not about bugs, is about implementing a product company-wide. It's more about long-term "trust" and "relations". Again don't get me wrong, I have big respect for you guys, but I personally don't know much about Proxmox, its employees (if any) and its leadership. I know I can trust Parallels, I know I can trust vmWare, I know I can trust Microsoft and Citrix. So when you - Proxmox - are playing "with the big guys" your users need to know if they can depend on you on the long run.

Not sure my comment will have any relevance, but I see alot of "trust" statements above...

I've lost count how many times MS has shutdown my xp and win7 machines because of my hardware hacking. My machines were legit licensed windows. True, shutdown is a harsh word and overkill but I really hate having to deal with calling MS helpline and putting in all those numbers, then having them read new codes back to me. Adobe, Kodak etal , are using internet based licensing I have a greater fear of the "big guys" pulling the plug than a group of open source communities. I deal with Kodak's license issues every week. But its what makes our business tick but it is very very annoying.

Anyway, after the last windows activation failure I personally moved on and not looked back.

 

[blackpaw]

As soon as RHEL7 beta is here, a new Proxmox VE kernel is also expected.

I'd ask, but I see from the web *everyone* is wondering when RHEL7 will be released End of this year was commonly mooted but no beta yet and getting a little late

I see from the roadmap you've majorly overhauled the backups - can you say what sort of changes? I was going to work on some hook scripts for my backups (mount/umount usb devices) - should I wait for now?

Spice Client as a complete replacement for the java vnc terminal - does this mean the Spice Client will be required to access the gui consoles?


And lastly, a big thanks for a great product, its literally making me sleep better at night, our XenServer cluster was giving me ednless hassles and I was logging in at the wee hours of the morning to scheudle reboots and check that the backups had worked (often they failed). ProxMox has been 100% reliable. Once we get out of testing I'll bug the boss for a commercial subscription.

Thanks - Lindsay

 

[mir]

Just a story about Oracle. Whe still use Oracle 9.2 which is still supported. 3 Years ago I discovered a great bug in the core which when you hit it would cause a core dump of the database. The bug can be found hidden inside the aggregate functionality. Only response from Oracle was that if we still wants to have support for our databases (we have about 15 9.2 running critical production) we must not use SQL including aggregates!! How much SQL/PLSQL can you make without using aggregates?

 

[tom]

I'd ask, but I see from the web *everyone* is wondering when RHEL7 will be released End of this year was commonly mooted but no beta yet and getting a little late

I hope there is a first public beta of RHEL7 quite soon, maybe end of this year - who knows. As soon as this is here, our devs also start with tests here.

I see from the roadmap you've majorly overhauled the backups - can you say what sort of changes? I was going to work on some hook scripts for my backups (mount/umount usb devices) - should I wait for now?

VM backup is inside KVM since 2.3, so as we have 1.7 now, there are also quite some changes there, but not in the functionality for the end users. for all details, check sources.

Spice Client as a complete replacement for the java vnc terminal - does this mean the Spice Client will be required to access the gui consoles?

yes. but as long as a spice client is not available on all major platforms, the java console is also available - fully configurable, just use what you like. but spiceterm is cool, you will not miss the java version.

And lastly, a big thanks for a great product, its literally making me sleep better at night, our XenServer cluster was giving me ednless hassles and I was logging in at the wee hours of the morning to scheudle reboots and check that the backups had worked (often they failed). ProxMox has been 100% reliable. Once we get out of testing I'll bug the boss for a commercial subscription.

Thanks - Lindsay

 

[Razva]

I agree with all of you. But you all should agree with this: all these company, producing bad/good/mediocre software, continue to provide the software, continue to provide updates/upgrades and continue existing. For that they have my trust.

Offtopic: did anybody *really* thought that nobody can tap and locate their phones, read their emails and hear everything they are talking? For me is just a matter of "hey, we're kinda admitting this". Come on, you live in a better world than the real one.

 

[m.ardito]

once I thought this was the best music player around..

http://en.wikipedia.org/wiki/Winamp



Marco

 

[p3x-749]

But you all should agree with this: all these company, producing bad/good/mediocre software, continue to provide the software, continue to provide updates/upgrades and continue existing. For that they have my trust.

You are dead wrong, sorry.
Their goal is a business and this means growth.
Even if you think with buying a license means getting insurance for future development, you're dead wrong.
All the "BIG-Five" are shopping the world with vast amounts of cash, buying competitors (and their products) or good products to be integrated in their portfolio.
This all leads to consolidation and products will simply vanish....taken out of sales and support...yes, it will be announced with a plan/calendar, but there is *nothing* you can do about it.
Even an ESCROW agreement will not rescue your businiess from its dependancy on that product (and that is what you are afraid of), because you can not handle it.
I've just seen several business in despair, because the firm with the "O" announced end of life for one of their premium integration platforms.
Yes, there is a new product (maybe for some there is even a technical migration path), but some smaller companies simply cannot afford migration of several years of investment without spending a fortune...they *are* left alone by their former supplier

Where's the difference with PVE?
Well, even if the main developer goes away, someone actually *can* take over....and if no one does, well, at least this is not a downside of PVE..your'e in the same boat as with the big commercial guys.

 

[blackpaw]

Something to remember is this isn't just a oss project - the whole format and tools are oss and widely used. I can take a qcow2 image and use it directly witgh a multitude of products - virt-manager, RHEL, ovirt etc, and it will run unchanged with the same drivers. Just try doing that with XenServer, VMWare or HyperV.

I have more faith that kvm/qemu will be around and activiely maintained in the future than I do with XenServer et al.