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?:
how to achive on par readyness of systems with fully opensource projects ....
I think that this is a mistake. It's not that you make a commitment to nothing but open source. It's only that when two otherwise equal options are available at a technical level, that open source licensing presents a strong benefit. Going out to choose software solely because it is open does not make sense, only treating open as an always beneficial factor.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Xen usually is operated via XO/XAPI whcih I do not feel so different form closed solutions due to commit hisroty in XAPI.
But they are, in every way. Every way. Commit history has nothing to do with it at all. Do you think that closed software comes from a single vendor? Open doesn't imply more than one. Closed doesn't imply only one. The number of committing entities is in no way related to our discussion, or to openness.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
DevOps or bare metal restore with proper incremental/dedup would benefit here. But I'm not aware of it in the open.
Otherway around, DevOps is all open. Not for any reason but market pressure. But DevOps is always open just by its nature.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Our "previous" system, the one we are going to phase out this winter, has been operated by an external supplier which used Xen + XO for this purpose -> quick readiness (less costs for the customer).
It is works, why replace it? Your concerns around Citrix and single vendor commits are totally unfounded and must be disregarded. XAPI is irrelevant, XO is not made by Citrix in any way, not a single commit, and Xen is very broad. This is an open solution that you already know with tools you know that meets all of your requirements. Is it the best, that depends on your needs. But without knowing a reason that you don't want to keep it, it seems a logical choice.
If your only concern is in any way connected to Citrix and XAPI, then I'm sure sticking with it is the right choice.
-
With XO you get agentless backups, add Veeam for free for robust agent backups as well. Add DevOps techniques and you can slowly move your backups to faster backup and restore times.
-
To give some perspective... worrying about who makes XAPI as a factor for Xen is like worrying about commits to Bash when selecting Linux.
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
ok put this simply. I go open source because it has more benefits then freeware. So I pick projects which do not depend on single corporate devel groups. Assume XAPI is not this. at least for the sake of stats I've extracted from git.
you go on premise with KVM on CentOS OR with Xen on opensuse leap (I would not go on ubuntu or debian - that's another topic).
Then I have to administer it. all open source because it pays more than freeware. I will use virt-manager with libvirt. This is ok with KVM even live migration is there. but Xen?
it starts appearing a bit risky IMHO probably XenCenter is the solution here. ok we hit another problem with XenCenter. just skip it.What about open source backup for VMs? To my knowledge you can eihter buy super big NASes for a longer retention policy of the OSes OR you backup the app (as I do in my web apps) and you simply try to make the OS backup irrelevant, AKA DevOps style a-la salt/ansible.
Otherwise you need baremetal-like restore of the OS. Which open source project does this?
Of course a proper mix of LVM snapshots, mount, rsnapshot (rsync) can do the work, but home made backup solution is probably NOT the way I would go in SMB (I did it with KVM just at LVM level, no dedup with rsnapshot - and retention was poor).
just link me to a proper quick to setup solution and honestly I will be able to sold benefits of opensource over freeware.
I miss this now. Then I will be able to sold openness even at hypervisor level not only application/OS level!Sadly the main reason why folks choose other than KVM is the above. The time and research and method and custom scripts is what KVM all about if you ask me. Just document your trial and errors and you will eventually have all this.
What I do, I have 2 KVM hosts running from centos minimal:
-
Check CPU support for Virtualization:
grep -E '(vmx|svm)' /proc/cpuinfo -
Install KVM:
yum groupinstall Virtualization "Virtualization Platform" "Virtualization Tools" -
Then make proper selinux and firewall configuration.
-
Then I load Fedora on my work laptop using virtualbox, and use virt-manager to manage the hosts.
-
Create passwordless SSH login between the hosts and preferably the virt-manager machine.
-
Alot of scripts and tools and guides to manage KVM and images got created by me due to going this route, for example I like to start the VM images with thin profile and expand over time, instead of allocating full amount of storage/Defalting the qcow2 image after time for backup or archiving (however I dont recommenced archvie cause qcow2 stores everything in 1 file/converting the qcow2 to other formats/ when will new image FVD be ready/ What VIRT IO drivers to install.... etc) The best thing is how simple is its to offline migrate VM images in KVM, and under Fedora there is tool like WinSCP called Gigolo I reckon, and with KVM you just need to copy the image disk and paste to the other Host and BAM, you copied the VM. due to there is no DRM feel or cause the file disk format is OSS everything is clear however with that comes the notion of everything can be setup in 100 diffrent ways.
If you want to bypass all this just get ESXi licensed, and your set.
-
-
@msff-amman-Itofficer said in open source hypervisors: do we really have them? do we really need them?:
If you want to bypass all this just get ESXi licensed, and your set.
Doing all that is easier than getting the license, I've tried.
If you want the power of KVM without the complexity, Scale HC3 is the way to go.
-
@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?:
Our "previous" system, the one we are going to phase out this winter, has been operated by an external supplier which used Xen + XO for this purpose -> quick readiness (less costs for the customer).
It is works, why replace it?
No I'm not replacing because of the current discussion. Totally unrelated. Presented here to explain why a lot of people in SMB with shortage of time and money do the XO way.
We replace because we change the ERP provider (at the time the provider was also the sysadmin) and we have gone a different route with HW and HV. Technically speacking I would had gone XenServer + XO from a tech perspective. Hyper-V has been the choice for other reasons.
OK Xen base is good.
Fact is: if a source is released but no one cares to keep it available outside of the company developing it, there is small chance it will be really useful for other than the company itself. So I do not see much value in its openness.
KVM could be a bit better but still lacks some tools.
Maybe I should re-state saying that yes core HV code is committed by enough etherogeneous sources that for (almost) sure it will survive any company both in KVM and Xen.
But most of the gear used to deploy on premise solutions in house (not vendor solutions) still is lacking. In Xen management tools are really limited: ok there is cmd line, but hey... seriously... SMB: really some SMB one-man-show writes tons of scripts from cmd line to manage everything?
KVM is ok at management level but backups are terrible. Xen could be better but most of the way incrementals are done is via XAPI, ASAP. Actually "bare metal" (maybe virtual metal?) restores would make it more viable as solution.
But why suggest them rather the Hyper-V if only the core is really developed in the open and not "just" released in the open? then you still need other tools - not open ones. At the end of the day: Yes we have open HV but don't know if we really need it.
(ok, I know we need but still I don't see the selling point other then I trust more open source then closed source. No one buys becaouse I have more trust: it is business not religion).
-
@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?:
Our "previous" system, the one we are going to phase out this winter, has been operated by an external supplier which used Xen + XO for this purpose -> quick readiness (less costs for the customer).
It is works, why replace it?
HV. Technically speacking I would had gone XenServer + XO from a tech perspective. Hyper-V has been the choice for other reasons.OK Xen base is good.
KVM is ok at management level but backups are terrible. Xen could be better but most of the way incrementals are done is via XAPI, ASAP. Actually "bare metal" (maybe virtual metal?) restores would make it more viable as solution.
Why do you care that much about the hypervisor backup capability? In my experience, agent backup permits cross-environment restore and is at least as fast and often more space efficient than the hypervisor-based backup.
It don't depends on the underlying hypervisor and often is offered for free… Linux has relax-and-recover for baremetal restore and a lot of tools for standard backups, namely rsnapshot, urbackup, the glorious bacula, attic, obnam, borg, pcbackup etc. I know there are also many tools for windows, maybe not oss but free like veeam.
Sometimes I used and hybrid approach that has proven to be very effective: backup the whole VM with a dumb system like whole machine export once a month, and backup just the data.
This way I can restore the full-blown machine that usually change very little apart of the data, and push the fresh data after the restore.I found this strategy very resource and space effective, and it can be executed with open source or at least free tools in any environment that I'm aware of.
-
Then entire discussion is whacked. But this statement...
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
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.
Reading code? how much of your employers time are your wasting with personal interest? If I caught an IT employee doing this they would be disciplined, and eventually terminated for continuing to do so.
Are you a trained software developer? Are you paid to inspect code? What the ever living hell do you think gives you the right to screw your employer over like that?
-
Just want to make sure I'm following this correctly. Is this the "Xen" that you guys are referring to ? https://www.xenproject.org/
If yes, what "GUI's" are available to manage Xen?
-
@FATeknollogee said in open source hypervisors: do we really have them? do we really need them?:
Just want to make sure I'm following this correctly. is this the "Xen" that you guys are referring to ? https://www.xenproject.org/
If yes, what "GUI's" are available to manage Xen?
Xen is the parent to XenServer. Generally when people speak of Xen they refer to XenServer. If they specify XAPI they are absolutely referring to XenServer as XenServer is the only system that uses the XAPI toolstack.
-
@FATeknollogee said in open source hypervisors: do we really have them? do we really need them?:
Just want to make sure I'm following this correctly. Is this the "Xen" that you guys are referring to ? https://www.xenproject.org/
If yes, what "GUI's" are available to manage Xen?
When using XenServer Xen Orchestra is the default
-
@DustinB3403 I am not talking about XenServer, only about Xen.
@scottalanmiller says Xen is viable in the SMB space, that's why I asked those questions -
@FATeknollogee said in open source hypervisors: do we really have them? do we really need them?:
@DustinB3403 I am not talking about XenServer, only about Xen.
@scottalanmiller says Xen is viable in the SMB space, that's why I asked those questionsHe means XenServer, very very very few SMB's would be implementing a Xen installation.
-
@scottalanmiller said in open source hypervisors: do we really have them? do we really need them?:
But you simply are not. Nothing binds you to it. I don't know what creates this perception, but Xen is totally viable without XS or XAPI. Just because recently people are talking about XO a lot so XAPI gets brought up doesn't tell us that Xen is only viable with it or that you are bound to it or anything like that.
@DustinB3403 I believe @scottalanmiller was talking about Xen not XenServer.
-
@FATeknollogee 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?:
But you simply are not. Nothing binds you to it. I don't know what creates this perception, but Xen is totally viable without XS or XAPI. Just because recently people are talking about XO a lot so XAPI gets brought up doesn't tell us that Xen is only viable with it or that you are bound to it or anything like that.
@DustinB3403 I believe @scottalanmiller was talking about Xen not XS.
OK think of it like this.
Xen is a tool set that you can make do what you want, but to get it to do something, you have to assemble a room.
XenServer is the toolset, the room, and everything else.
Xen is used by Amazon and company, it is very powerful, but it isn't for the SMB space. Not like Hyper-V and XenServer are.
-
@JaredBusch said in open source hypervisors: do we really have them? do we really need them?:
Then entire discussion is whacked. But this statement...
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
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.
Reading code? how much of your employers time are your wasting with personal interest? If I caught an IT employee doing this they would be disciplined, and eventually terminated for continuing to do so.
Are you a trained software developer? Are you paid to inspect code? What the ever living hell do you think gives you the right to screw your employer over like that?
10 years of code devel. I do this out of my working time. I inspect code in working time only when paid for it
-
@matteo-nunziati said in open source hypervisors: do we really have them? do we really need them?:
Fact is: if a source is released but no one cares to keep it available outside of the company developing it, there is small chance it will be really useful for other than the company itself. So I do not see much value in its openness.
Right, and we know that there is zero chance of this with Xen and KVM. Zero. So openness is 100% effective here for the one thing you consider to be its value (which is not what openness is about.) So in your skewed case of redefining the value, Xen and KVM meet your desires as perfectly as it is possible to do. The penultimate definition of open to your standards.
