I swapped over my home lab from ESXi (VMUG) to PM a few weeks ago. Everything is set up and working without issue, save one. The speed of my VM's is dramatically slower than under ESXi. This surprised me as PM is supposed to be more efficient.
Hardware:
I have a pair of HPE ProLiant MicroServer Gen10 Plus
Xeon E-2224 CPU (1 x 4 core)
64GB RAM
4 x 1TB SSD (ZFS) - Under ESXi, this was RAID 5. VM's are stored locally.
After the swap over I replaced my RAID card with 10GB 2 x SFP+ card. Was 4 x 1GB networking, now 2 x 10GB. Saw the same issue before and after.
I have four active VM's currently. Two Windows 2019 servers (DC and DC backup), Win 10 Enterprise and Linux. One DC and one other VM per host. They are the same vm's I had under ESXi. Same 6GB RAM each, CPU allotment, HDD, across the board. Since the swap I have noticed the speed of the VM's is 30-60% slower. Takes way longer to apply Windows updates, reboots are "have a coffee then check back" slow. Under ESX the vm were never powerhouses, but it performed better. Latest VirtIO drivers are installed. No errors in logs, no crashes or issues save the slow vm.
I set vm CPU to "host" (default cpu flags) Tried both QEMU64 and KVM64 but see no real performance difference.
Thinking it may be a vm conversion issue, I rebuilt the Win 10 Enterprise vm from scratch and see the same performance hit. Included some info below. Let me know if you need anything else.
root@PROXMOX1:~# pveversion
pve-manager/8.2.4/faa83925c9641325 (running kernel: 6.8.8-3-pve)
root@PROXMOX1:~# zpool status
pool: PROXMOX1_Local
state: ONLINE
scan: scrub repaired 0B in 00:02:43 with 0 errors on Sun Aug 11 00:26:44 2024
config:
NAME STATE READ WRITE CKSUM
PROXMOX1_Local ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
wwn-0x500a0751e4e17270-part1 ONLINE 0 0 0
wwn-0x5002538e7031aec5-part1 ONLINE 0 0 0
wwn-0x5002538e7031aebc-part1 ONLINE 0 0 0
wwn-0x5002538e7031a769-part1 ONLINE 0 0 0
errors: No known data errors
root@PROXMOX1:~# free -h
total used free shared buff/cache available
Mem: 62Gi 40Gi 1.7Gi 66Mi 21Gi 22Gi
Swap: 8.0Gi 256Ki 8.0Gi
root@PROXMOX1:~# pveperf
CPU BOGOMIPS: 27199.24
REGEX/SECOND: 4967018
HD SIZE: 67.73 GB (/dev/mapper/pve-root)
BUFFERED READS: 400.33 MB/sec
AVERAGE SEEK TIME: 0.17 ms
FSYNCS/SECOND: 178.50
DNS EXT: 376.45 ms
DNS INT: 334.74 ms (domain name deleted)
root@PROXMOX1:~# qm config 100 (Win 10 Enterprise)
agent: 1
bios: seabios
boot: order=scsi0;scsi1
cores: 4
cpu: host
machine: pc-q35-8.1,viommu=intel
memory: 6144
meta: creation-qemu=8.1.5,ctime=1717263695
name: Win10Ent
net0: virtio=00:50:56:9f:b6:67,bridge=vmbr0
numa: 0
ostype: win10
sata0: none,media=cdrom
scsi0: PROXMOX1_Local:vm-100-disk-0,format=raw,size=130G
scsi1: PROXMOX1_Local:vm-100-disk-1,format=raw,size=75G
scsihw: virtio-scsi-single
smbios1: uuid=42033965-ef82-38c0-44c9-21e87639e8ef
sockets: 1
vmgenid: 1f48662e-0a9b-4228-bf5b-028ae48f3008
Hardware:
I have a pair of HPE ProLiant MicroServer Gen10 Plus
Xeon E-2224 CPU (1 x 4 core)
64GB RAM
4 x 1TB SSD (ZFS) - Under ESXi, this was RAID 5. VM's are stored locally.
After the swap over I replaced my RAID card with 10GB 2 x SFP+ card. Was 4 x 1GB networking, now 2 x 10GB. Saw the same issue before and after.
I have four active VM's currently. Two Windows 2019 servers (DC and DC backup), Win 10 Enterprise and Linux. One DC and one other VM per host. They are the same vm's I had under ESXi. Same 6GB RAM each, CPU allotment, HDD, across the board. Since the swap I have noticed the speed of the VM's is 30-60% slower. Takes way longer to apply Windows updates, reboots are "have a coffee then check back" slow. Under ESX the vm were never powerhouses, but it performed better. Latest VirtIO drivers are installed. No errors in logs, no crashes or issues save the slow vm.
I set vm CPU to "host" (default cpu flags) Tried both QEMU64 and KVM64 but see no real performance difference.
Thinking it may be a vm conversion issue, I rebuilt the Win 10 Enterprise vm from scratch and see the same performance hit. Included some info below. Let me know if you need anything else.
root@PROXMOX1:~# pveversion
pve-manager/8.2.4/faa83925c9641325 (running kernel: 6.8.8-3-pve)
root@PROXMOX1:~# zpool status
pool: PROXMOX1_Local
state: ONLINE
scan: scrub repaired 0B in 00:02:43 with 0 errors on Sun Aug 11 00:26:44 2024
config:
NAME STATE READ WRITE CKSUM
PROXMOX1_Local ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
wwn-0x500a0751e4e17270-part1 ONLINE 0 0 0
wwn-0x5002538e7031aec5-part1 ONLINE 0 0 0
wwn-0x5002538e7031aebc-part1 ONLINE 0 0 0
wwn-0x5002538e7031a769-part1 ONLINE 0 0 0
errors: No known data errors
root@PROXMOX1:~# free -h
total used free shared buff/cache available
Mem: 62Gi 40Gi 1.7Gi 66Mi 21Gi 22Gi
Swap: 8.0Gi 256Ki 8.0Gi
root@PROXMOX1:~# pveperf
CPU BOGOMIPS: 27199.24
REGEX/SECOND: 4967018
HD SIZE: 67.73 GB (/dev/mapper/pve-root)
BUFFERED READS: 400.33 MB/sec
AVERAGE SEEK TIME: 0.17 ms
FSYNCS/SECOND: 178.50
DNS EXT: 376.45 ms
DNS INT: 334.74 ms (domain name deleted)
root@PROXMOX1:~# qm config 100 (Win 10 Enterprise)
agent: 1
bios: seabios
boot: order=scsi0;scsi1
cores: 4
cpu: host
machine: pc-q35-8.1,viommu=intel
memory: 6144
meta: creation-qemu=8.1.5,ctime=1717263695
name: Win10Ent
net0: virtio=00:50:56:9f:b6:67,bridge=vmbr0
numa: 0
ostype: win10
sata0: none,media=cdrom
scsi0: PROXMOX1_Local:vm-100-disk-0,format=raw,size=130G
scsi1: PROXMOX1_Local:vm-100-disk-1,format=raw,size=75G
scsihw: virtio-scsi-single
smbios1: uuid=42033965-ef82-38c0-44c9-21e87639e8ef
sockets: 1
vmgenid: 1f48662e-0a9b-4228-bf5b-028ae48f3008
