QEMU/KVM + Ceph Librbd Performance tuning

[Whatever]

With respect to blog at ceph.io
https://ceph.io/en/news/blog/2022/qemu-kvm-tuning/

Which memory allocator and libRBD are used in Proxmox?
Are the suggested optimizations in article above suitable for PVE+CEPH optimization?

 

[aaron]

Hmm, I gave it a quick glance and a lot of these things are either default or can be easily done. No cache for the disk drive for example, or using IO threading for each disk image of the VM.

The librbd linked by qemu:

root@cephtest1:~# ldd /usr/bin/kvm  | grep librbd
    librbd.so.1 => /lib/librbd.so.1 (0x00007f89397a3000

depends on the installed version. On my test system that is currently running Pacific (16):

root@cephtest1:~# ls -la /lib/librbd.so.1
lrwxrwxrwx 1 root root 16 Oct 19 07:27 /lib/librbd.so.1 -> librbd.so.1.16.0

The allocator could be one thing that could improve the performance. But I cannot answer that right away.

 

[Whatever]

The allocator could be one thing that could improve the performance. But I cannot answer that right away.

Just wondering whether changing allocator would have an impact on QEMU performance or not?