Troubleshooting › Backup Read Throughput is Less When the Backup Node is RHEL, CentOS, or Oracle Linux 5.x On a Xen- based PVM

Backup Read Throughput is Less When the Backup Node is RHEL, CentOS, or Oracle Linux 5.x On a Xen- based PVM

Valid on Red Hat Enterprise Linux (RHEL), CentOS, SUSE Linux Enterprise Server (SLES) 11.x, and Oracle Linux Server 6.x on a Xen-based PVM

Symptom

When I back up an RHEL, CentOS, SLES, Oracle Linux 6.x server on a Xen-based PVM, the backup read throughput value is less.

Solution

To resolve this issue, configure the disk input/output (I/O) scheduler of the VM.

Follow these steps:

1. Open the Command Prompt window on the backup source node.
2. Run the following command to identify the number of disks the VM has:

   ls /dev/xvd*
   

3. Run the following command to identify the I/O scheduler algorithm of the disks:

   cat /sys/block/xvda/queue/scheduler
   

   The following result is displayed on the VM:

   [noop] anticipatory deadline cfq
   

4. Run the following command to change the I/O scheduler to cfq:

   echo "cfq" > /sys/block/xvda/queue/scheduler
   

5. Verify the I/O scheduler again.

   The following result is displayed on the VM:

   noop anticipatory deadline [cfq]
   

6. Change the I/O scheduler for every disk.
7. Rerun the backup job.

The backup job runs with an optimum backup read throughput.