Hello
We have 4 servers with Proxmox using local hard drives (even in version 2.3 but at the moment it is not important).
To improve the performance of virtual machines purchased a disk array EMC VNX 5200 - block access by card 8Gb FC (lpe12002 with 2 x MiniGBIC FC 8Gb / s). 25 x 600GB 10krpm (4 x disk with RAID5 and 16 x drive with RAID 10). Each server is connected to the array 2 cables (on EMC Failover Mode: Active-Active mode (ALUA) -failovermode 4).
I devoted for testing one server (IBM System x3400) and I am shocked Proxmox 3.3 performance compared to the performance of Hyper-v 2012r2. Of course, in the case of Proxmox configured the multipath (Alua) and hyper-v MPIO.
Below screenshots of the test read / write HDTune program running on a virtual machine with a clean newly installed system Windows2008r2 - without any updates. Tests hyper-v and Proxmox done on the same physical server.
If HyperV 2012r2 on two resources (RAID5 and RAID10) I created NTFS partitions.
In the case of Proxmox 3.3 obtain similar performance with LVM as well as a mounted partition with ext3 and raw disks (writeback cache). Similar perfomance is with cache writeback an cache none. I tried virtio driver and have this same performance like without driver virtio.
We use Proxmox without purchased support because we are a public institution (Public Library), which has no additional funds for an annual subscription fee - array we bought the resources acquired of the Ministry of Culture.
I would like to stay with Proxmox, but if I can not improve the performance of this I will be forced to change virtualizer.
If I have to perform additional tests (indifferently on which hypervisor) or share some logs please let us know.
We have 4 servers with Proxmox using local hard drives (even in version 2.3 but at the moment it is not important).
To improve the performance of virtual machines purchased a disk array EMC VNX 5200 - block access by card 8Gb FC (lpe12002 with 2 x MiniGBIC FC 8Gb / s). 25 x 600GB 10krpm (4 x disk with RAID5 and 16 x drive with RAID 10). Each server is connected to the array 2 cables (on EMC Failover Mode: Active-Active mode (ALUA) -failovermode 4).
I devoted for testing one server (IBM System x3400) and I am shocked Proxmox 3.3 performance compared to the performance of Hyper-v 2012r2. Of course, in the case of Proxmox configured the multipath (Alua) and hyper-v MPIO.
Below screenshots of the test read / write HDTune program running on a virtual machine with a clean newly installed system Windows2008r2 - without any updates. Tests hyper-v and Proxmox done on the same physical server.
If HyperV 2012r2 on two resources (RAID5 and RAID10) I created NTFS partitions.
In the case of Proxmox 3.3 obtain similar performance with LVM as well as a mounted partition with ext3 and raw disks (writeback cache). Similar perfomance is with cache writeback an cache none. I tried virtio driver and have this same performance like without driver virtio.
We use Proxmox without purchased support because we are a public institution (Public Library), which has no additional funds for an annual subscription fee - array we bought the resources acquired of the Ministry of Culture.
I would like to stay with Proxmox, but if I can not improve the performance of this I will be forced to change virtualizer.
If I have to perform additional tests (indifferently on which hypervisor) or share some logs please let us know.