New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
I guest it will allow the host to grab memory from guest. This blog post explains it more clearly.
I don't see why a host would need to KVM balloon
Did your host (in your plan) give you separate dedicated/burst RAM limits? If not, your server node is probably oversold.
@tsc burst ram is not part of kvm
@miTgiB http://www.linux-kvm.com/content/memory-ballooning-feature-coming-soon-kvm
While not in general use today, it seems plausible considering the feature was introduced almost three years ago.
@tsc You refer to memory ballooning, yet talked about burst ram, two different things, burst ram is an OpenVZ thing, memory ballooning is a xen/kvm thing.
@miTgiB I guess I was attempting to use burst RAM as a general term. My mistake.
This one is a BuyVM $25/y 128 kvm, actually it's performing well and I trust Fran that he will not do something like N****c. I guess this is probably the virtio thing, do you have such thing on your server, @miTgiB?
I'm not sure if the 2.6.18 kernel even supports ballooning, so we wouldn't be able to use it on the HN even if we wanted to
With that being said, we do limited stock on all plans to keep from any sort of overselling. We max out on virtual cores before we do any other resources. We have like 1.5TB spare on each physical box even after everyone being allocated 100% of their space by LVM, so we'll likely put a couple storage plans on each box to make it even better
Thanks for trying us out!
CentOS 5.6 HN 2.6.18-238.9.1.el5
Debian 5 VM
CentOS 5.6 VM
It's simply there, for the very valid reason of over committing memory, now, how the admin manages that over commitment is the real dirt in the details question. I simply had no idea how KVM would react to memory over commitment in real world use, so oversold a single KVM node to 17gb while it has 16gb of physical ram, and while it had 7gb of swap in use, there was no IO starvation, no complaints of any kind about VPS performance, and helped in making my choice that KVM was safe to sell to 15.5gb on a 16gb node, where Xen I didn't feel safe committing that much ram to VM's. With a fully loaded node, I am seeing about 30% CPU usage on average. Once 8gb sticks are released in larger quantity at affordable pricing, the E3-1200 series will defiantly benefit and still be a very viable option for KVM nodes with 32gb without taxing the CPU, if it needs a 4 or 8 drive raid will be the question, but I think some of the sketchier providers doing 2 drive raid1 with huge amounts of ram, you'll see the results when you see KVM at OpenVZ pricing, and hopefully you'll run like hell.
The 2.6.18 kernel tree also has some basic KSM support for KVM, which I've played with for deduping page files, but I was only testing it with my X6 node and found it added about 5% load with little to no benefit with a node full of linux and FreeBSD vm's, but for a node full of Windows VM's it would probably be a great util. I do have a 2.6.32 based KVM node, but have not done anything with KSM yet on it.
KVM balooning is a manual process (the admin has to manually cause each guest to balloon memory up/down, unlike on eg: Xen PV where balooning is automaticly done as the host server runs out of memory)
It's far more likely KVM hosts will overcommit memory with KSM (which is automatic) instead of balooning.
KSM > Balooning, both in terms of ease of deployment, as well as due to the fact the client's available memory doesn't actually decrease. KSM is like gzip compression for memory pages.
Here is how to have virtio working: http://www.linux-kvm.org/page/Virtio
"Virtio balloon driver" seems to be part of it. And proxmox.com says: "Modern Linux Kernels does include the Balloon drivers by default. It works out of the box, and you only need to set the VM to "Automatically allocate memory within this range".
Please is there anyone who tried this ballooning feature and found it useful for the case of Linux VPSs or how it compares to KSM (Kernel Samepage Merging)?
PS: is this ballooning feature automated now, how can be configured? (http://www.linux-kvm.org/page/Projects/auto-ballooning)
thread is by far outdated.
Of course menwhile a hypervisor should be able to do this automatically and probably a very high percentage of all providers use it.
imho it does not need to compare to ksm as it simply adds another approach to leverage ressources
Sure it is, it was from 2011! No idea why it's revived today, but it's good to see good providers still here all these years.