Open Source Hypervisors: do we really have them? do we really need them?
-
@Francesco-Provino said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati Xen can be fully used without XAPI, that are also an OSS project.
XenServer is the Xen package with XAPI, but any enterprise distro provide Xen WITHOUTH XAPI, namely SuSe and Ubuntu. You can use it effectively with libvirt or the xl toolstack.At the moment I don't see any risk associated with Xen being dropped by anyone, because it's the widely used hypervisor in the world. Almost any public cloud use that. Just the fact that AWS is built on Xen it's a guarantee that it cannot became abandonware in any way. Amazon alone could support the entire Xen development with 0.1% the revenues from the AWS cloud. They have all the interest in maintain Xen healthy.
And IBM's SoftLayer and Rackspace (which was important until this morning) are both Xen as well. Neither is on par with Amazon, but both are large enough to support it without any other vendors involved.
And OpenStack is heavily involved with Xen (sans XAPI). Not as much as with KVM today, but is a major player in the private cloud space with Xen.
-
@matteo-nunziati here is a Xen wiki link about the Xen toolstack that you can use to manage Xen.
-
Don't forget that things like the open source and multi-vendor approach to KVM is what allows us to have stuff like Scale HC3, as well!
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Of course we have both KVM and Xen but let think about them a little more...
And BHyve
type-2?
No, type 1.
ok, type 1 for everyone but some of their presentations back in 2012 X-D! They classify KVM as type-2 too...
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Of course we have both KVM and Xen but let think about them a little more...
And BHyve
type-2?
No, type 1.
ok, type 1 for everyone but some of their presentations back in 2012 X-D! They classify KVM as type-2 too...
Whose presentations?
-
Whenever Xen is questioned, why only bring Amazon? Are there any SMB examples that is using Xen or XenServer without a doubt?
If it wasn't for XenSever, I could care less about using Xen.
As for KVM, I believe there are more community backing than Xen.
-
@black3dynamite said in open source hypervisors: do we really have them? do we really need them?:
Whenever Xen is questioned, why only bring Amazon? Are there any SMB examples that is using Xen or XenServer without a doubt?
Can you name SMB examples of anything? We all know that they are out there, but SMB aren't useful examples for anything. It's the nature of the SMB. I mean we know that there are a lot of Xen users here in ML, how many examples do you need?
We know that SMB does all kinds of things, we see them all of the time. But SMB examples are dangerous for lots of reasons. For one, they are often unknown (we don't have any reason to think that what they do is good.) For another, they are not public (no one talks about what they do.)
Even if no SMB uses Xen, doesn't mean that it's bad, only that SMBs aren't doing it. Amazon is an important example here because no single SMB is going to fund any project like this, but Amazon will. Amazon is also important because they run the fastest most secure environment on the planet (according to the CIA), as well as the largest. Their expertise carries rather significant value. As does their cost analysis.
We can learn a lot from one Amazon that we can't learn from ten thousand SMBs.
-
@black3dynamite said in open source hypervisors: do we really have them? do we really need them?:
If it wasn't for XenSever, I could care less about using Xen.
Why is that? My experience is mostly the opposite, only found XS interesting recently and the maintenance of it made me essentially not care anymore. It's Xen that is really interesting. It's more advanced and has some VMware level features, like full fault tolerance that are removed in XS.
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
XAPI is so non standard wrt Xen Project source code, that no one is packaging it and current Citrix efforts are focused only on Centos because they use it for Dom 0.
On their wiki they say that the package for Ubuntu. Maybe this has changed, but given the nature of XCP and XAPI, would you want it anywhere but CentOS? Is this in any way a negative?
No CentOS is ok.
but the fact is that commit history is strictly dominated by citrix and in all those years no one has cared / been able( don't know) to make it compile on any distribution as you do with KVM or libvirt or so.
Basically if you want to go with Xen world you have to go XenServer + XO (or XenServer + Unitrends, but don't know if they use XAPI). outside of XenServer w/ XAPI there is not choice just a fixed suboptimal management path (libvirt)
Here in these days seems that open source alternatives for Xen simply means: stay away from Xen go KVM. And this is cast in stone since years now. Not really nice.Ubuntu is tagged as experimental and still compiles on 14.04 only.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
but the fact is that commit history is strictly dominated by citrix and in all those years no one has cared / been able( don't know) to make it compile on any distribution as you do with KVM or libvirt or so.
The big piece is... no one cared. It's a distro system. What would be the reason for someone to want to port it to something else? It's not like KVM, it's a totally different animal. KVM is part of the kernel. XCP and XS are distributions.
I don't see the slightest concern there, just totally different community and market pressures. It makes total sense for KVM to be everywhere. It makes really no sense for XCP and XAPI to be anywhere but where it is.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Basically if you want to go with Xen world you have to go XenServer ....
Oh no, not at all. that's not even the common way.
-
I think that the issue here is feeling that XAPI is more important than it is. If you stop looking at XAPI and only look at Xen, suddenly Xen is more powerful, more flexible, available on lots of platforms, has a way longer history than Vmware, etc.
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
My opinion is as far from this as possible. Open source is specifically protection against this. Nothing carries this risk more than closed source. Open source is, in fact, the sole protection against this problem. What you describe is literally the key risk of closed source software, not open source.
Yes I agree on a theoretical level, but how much of this is really applicable to Xen and KVM? Look at other things like upstart. Yes Canonical is way more "unstable" than IBM or Red Hat, but this is an example of how loosing the main sponsor usually means the projects start lagging and slowing down. Something which doesn't happen if the code is committed by etherogeneous subjects.
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Basically if you want to go with Xen world you have to go XenServer ....
Oh no, not at all. that's not even the common way.
limit this to SMB, don't think about larger groups. I know Amazon is by no means a XenServer user
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Of course we have both KVM and Xen but let think about them a little more...
And BHyve
type-2?
No, type 1.
ok, type 1 for everyone but some of their presentations back in 2012 X-D! They classify KVM as type-2 too...
Whose presentations?
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Basically if you want to go with Xen world you have to go XenServer ....
Oh no, not at all. that's not even the common way.
limit this to SMB, don't think about larger groups. I know Amazon is by no means a XenServer user
SMBs recently have started toying with XS because it had a brief moment in the sun. I come from SMBs that used Xen without XS, it is the environment I've seen the most in the SMB (before Jared says anything, I'm not suggesting it is the most popular, I'm just saying I've seen it a lot.)
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
@black3dynamite said in open source hypervisors: do we really have them? do we really need them?:
If it wasn't for XenSever, I could care less about using Xen.
Why is that? My experience is mostly the opposite, only found XS interesting recently and the maintenance of it made me essentially not care anymore. It's Xen that is really interesting. It's more advanced and has some VMware level features, like full fault tolerance that are removed in XS.
Because it was less of pain to get setup compare to my experience with Xen.
-
@black3dynamite said in open source hypervisors: do we really have them? do we really need them?:
If it wasn't for XenSever, I could care less about using Xen.
tht's my point in SMB, at least for the limited exposure I've seen about Xen in SMB. They go the XO + XS way, but to me this would mean just bind myself to XAPI, and I see this as a Citrix project no one else cares about. So what difference with Hyper-V?
Disclosure: I really like open source and looking at code for simplier projects really saved my day, just trying to remove some doubts of mine about KVM and Xen.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
My opinion is as far from this as possible. Open source is specifically protection against this. Nothing carries this risk more than closed source. Open source is, in fact, the sole protection against this problem. What you describe is literally the key risk of closed source software, not open source.
Yes I agree on a theoretical level, but how much of this is really applicable to Xen and KVM?
Completely and utterly. About as much as it has ever applies to any project in history. Linux itself would be protected more, but little else. These are actually the prototypical examples of how dramatically this would protect us.
-
trying to stats KVM git repo... my pc is burning...
