Amazon AWS Leaving Xen for KVM
-
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
This is enormous news. This, more than anything, signals the end of the road for Xen. A very sad day.
Oh man I only just really started using XEN better get that test server racked and KVM installed lol
-
@hobbit666 said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
This is enormous news. This, more than anything, signals the end of the road for Xen. A very sad day.
Oh man I only just really started using XEN better get that test server racked and KVM installed lol
Now is the chance to get ahead of the curve rather than behind it
-
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
If Hyper-V was Windows, it wouldn't need Windows in the Dom0. It's specifically that it isn't that that is required.
Now I'm lost - Hyper-V still has a Dom0 even when installed as pure Hyper-V?
You never question why for a Hyper-V Core required so much damn install space? The Management VM (DOM0) just runs headless.
-
@storageninja said in Amazon AWS Leaving Xen for KVM:
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
If Hyper-V was Windows, it wouldn't need Windows in the Dom0. It's specifically that it isn't that that is required.
Now I'm lost - Hyper-V still has a Dom0 even when installed as pure Hyper-V?
You never question why for a Hyper-V Core required so much damn install space? The Management VM (DOM0) just runs headless.
TBH if you look at really small partitions maybe Hyper-V requires one of the biggest (but not sure about KVM), anyway my hyper-v install is around 26GB. not a big deal .
-
Yeah I can't imagine an SMB who cares about <30gb for a hypervisor.
-
@tim_g said in Amazon AWS Leaving Xen for KVM:
Yeah I can't imagine an SMB who cares about >30gb for a hypervisor.
Do you mean <30GB? Because at some point larger than 30GB, everyone cares. What point, I don't know, but at some point, for sure.
-
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Yeah I can't imagine an SMB who cares about >30gb for a hypervisor.
Do you mean <30GB? Because at some point larger than 30GB, everyone cares. What point, I don't know, but at some point, for sure.
Yeah it was supposed to be less than, not greater.
-
@matteo-nunziati said in Amazon AWS Leaving Xen for KVM:
@storageninja said in Amazon AWS Leaving Xen for KVM:
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
If Hyper-V was Windows, it wouldn't need Windows in the Dom0. It's specifically that it isn't that that is required.
Now I'm lost - Hyper-V still has a Dom0 even when installed as pure Hyper-V?
You never question why for a Hyper-V Core required so much damn install space? The Management VM (DOM0) just runs headless.
TBH if you look at really small partitions maybe Hyper-V requires one of the biggest (but not sure about KVM), anyway my hyper-v install is around 26GB. not a big deal .
Reinstalled in Feb, so there have been some extras added. Plus it was installed with the hypervisor role from the ISO, not minimal and added KVM. So there are a few extras that aren't needed like gluster client, openscap, etc.
But OpenSCAP is helpful on the host because you can scan your VMs directly from the host without the software being in the VM.
-
@stacksofplates said in Amazon AWS Leaving Xen for KVM:
@matteo-nunziati said in Amazon AWS Leaving Xen for KVM:
@storageninja said in Amazon AWS Leaving Xen for KVM:
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
If Hyper-V was Windows, it wouldn't need Windows in the Dom0. It's specifically that it isn't that that is required.
Now I'm lost - Hyper-V still has a Dom0 even when installed as pure Hyper-V?
You never question why for a Hyper-V Core required so much damn install space? The Management VM (DOM0) just runs headless.
TBH if you look at really small partitions maybe Hyper-V requires one of the biggest (but not sure about KVM), anyway my hyper-v install is around 26GB. not a big deal .
Reinstalled in Feb, so there have been some extras added. Plus it was installed with the hypervisor role from the ISO, not minimal and added KVM. So there are a few extras that aren't needed like gluster client, openscap, etc.
But OpenSCAP is helpful on the host because you can scan your VMs directly from the host without the software being in the VM.
Group install virtualization doesn’t install they same packages?
-
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Yeah I can't imagine an SMB who cares about >30gb for a hypervisor.
Do you mean <30GB? Because at some point larger than 30GB, everyone cares. What point, I don't know, but at some point, for sure.
It matters from a secondary standpoint of..
- That's a lot of damn code. There are security implications of having that much to keep patched (more patch windows, more attack points).
- If you do PXE/AutoDeploy scenarios it slows down your provisioning/boot time. (Can you even run Hyper-V in a supported, stateless PXE config?)
- Lack of optimizations for embedded installs. It's also a symptom of Hyper-V not being really designed for an embedded install (I know some Nutanix guys tried it but kept burning out SATA DOMs). This is slightly less of a concern with stuff like BOSS modules now an option on 14Gen servers (still that's ~$250 extra per host).
You can get KVM pretty lean, and the ESXi 6.5 ISO weighs in at 324MB (of that the hypervisor is a little less than 1/2 of, the rest being tools images).
-
@storageninja said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Yeah I can't imagine an SMB who cares about >30gb for a hypervisor.
Do you mean <30GB? Because at some point larger than 30GB, everyone cares. What point, I don't know, but at some point, for sure.
It matters from a secondary standpoint of..
- That's a lot of damn code. There are security implications of having that much to keep patched (more patch windows, more attack points).
- If you do PXE/AutoDeploy scenarios it slows down your provisioning/boot time. (Can you even run Hyper-V in a supported, stateless PXE config?)
- Lack of optimizations for embedded installs. It's also a symptom of Hyper-V not being really designed for an embedded install (I know some Nutanix guys tried it but kept burning out SATA DOMs). This is slightly less of a concern with stuff like BOSS modules now an option on 14Gen servers (still that's ~$250 extra per host).
You can get KVM pretty lean, and the ESXi 6.5 ISO weighs in at 324MB (of that the hypervisor is a little less than 1/2 of, the rest being tools images).
All good points, for sure. I'm not arguing them.
My whole point is that (this being an SMB forum) we don't deploy hundreds or thousands of Hypervisors via PXE. We don't care if the hypervisor is 30GB, like, ever. Patching is never a problem in SMB because patching is done automatically and you have scheduled down time for that. If you can't have downtime for scheduled maintenance, you would have some kind of HA... because then it would make sense due to your business not being ablet to afford down time.
Sure, in large datacenter warehouses where I can see your examples mattering. Definitely, every GB matters. When dealing with thousands of identical systems and constantly deploying, every little thing in every aspect matters. But again, SMB here... try to keep your points in the appropriate context.
-
@tim_g said in Amazon AWS Leaving Xen for KVM:
Patching is never a problem in SMB because patching is done automatically and you have scheduled down time for that.
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
-
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Patching is never a problem in SMB because patching is done automatically and you have scheduled down time for that.
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work nights or weekends.
-
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Patching is never a problem in SMB because patching is done automatically and you have scheduled down time for that.
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work
nights or weekends.FTFY
-
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Patching is never a problem in SMB because patching is done automatically and you have scheduled down time for that.
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work
nights or weekends.FTFY
LOL
-
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@dashrender said in Amazon AWS Leaving Xen for KVM:
@scottalanmiller said in Amazon AWS Leaving Xen for KVM:
@tim_g said in Amazon AWS Leaving Xen for KVM:
Patching is never a problem in SMB because patching is done automatically and you have scheduled down time for that.
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to
Hire Indians to do it for you at night.FTFY
FTFY.
-
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work nights or weekends.
If only there was a way to schedule updates to run outside of business hours.....
-
@r3dpand4 said in Amazon AWS Leaving Xen for KVM:
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work nights or weekends.
If only there was a way to schedule updates to run outside of business hours.....
One can dream. Maybe in a far off, distant IT future....
-
@r3dpand4 said in Amazon AWS Leaving Xen for KVM:
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work nights or weekends.
If only there was a way to schedule updates to run outside of business hours.....
When you have to manufacture in Asia, and trucks back up if they can't print labels at 3AM US time you stop having "outside of business hours". An increasing amount of (even Small business's) don't have clear gaps, and you need someone to be ready to "fix" things if that patching fails, or brings something down.
You can have monitoring systems that will trigger a TAS to page the on-call, but if that fails there is nothing worse than waking up at 7AM and discovering the entire office is dead in the water. Follow the sun operations are bleeding into more and more companies.
-
@storageninja said in Amazon AWS Leaving Xen for KVM:
@r3dpand4 said in Amazon AWS Leaving Xen for KVM:
not if you read SW, you'd think no SMB knows to patch and those that do can never get downtime for them.
That's because they don't want to work nights or weekends.
If only there was a way to schedule updates to run outside of business hours.....
When you have to manufacture in Asia, and trucks back up if they can't print labels at 3AM US time you stop having "outside of business hours". An increasing amount of (even Small business's) don't have clear gaps, and you need someone to be ready to "fix" things if that patching fails, or brings something down.
You can have monitoring systems that will trigger a TAS to page the on-call, but if that fails there is nothing worse than waking up at 7AM and discovering the entire office is dead in the water. Follow the sun operations are bleeding into more and more companies.
Ya this is why that stuff should be automated as much as possible. If something fails during a patch, have it nuke it and bring another one up. Obviously someone still needs to be ready but let the machines do what they were built to do.
