Dear forum members,
I am exporting zfs file system from FreeBSD 9.0 amd64 to VMware ESXi over NFS. On this datastore I create virtual disks which are presented to Windows 2008 R2 virtual machines as locally attached disks used for scheduled VSS backups.
When first backup is created, data is only written to disk, and I get satisfactory speed of ~10M/s, as read from zpool iostat 5. This creates 20Gb backup in around 30 minutes. However, each consecutive backup takes around 3 hours, and zpool iostat shows reads of ~500K and writes of ~20K.
What could be the reason for such terrible degradation, and how could i resolve it?
I am exporting zfs file system from FreeBSD 9.0 amd64 to VMware ESXi over NFS. On this datastore I create virtual disks which are presented to Windows 2008 R2 virtual machines as locally attached disks used for scheduled VSS backups.
When first backup is created, data is only written to disk, and I get satisfactory speed of ~10M/s, as read from zpool iostat 5. This creates 20Gb backup in around 30 minutes. However, each consecutive backup takes around 3 hours, and zpool iostat shows reads of ~500K and writes of ~20K.
What could be the reason for such terrible degradation, and how could i resolve it?