Backups slow on 1 File server in ESX4 cluster
I was wondering if someone had any advice with an issue I are having with backups of 2files servers running as Virtual Machines. Both machines are VM’s in the ESX4 cluster and are configured as follows:
2003 server - 4 vCPU's and 4gb RAM, it is a 32bit server with the flexible network card in VMware.
2008 server 1vCPU 64bit server with 3gb RAM with the E1000 network card in VMware.
1 file server is a Windows 2008 R2 Enterprise server which has approx 450gb of data and
the other file server is a Windows 2003 R2 Enterprise Server which has approx 450gb of data.
To do a full backup on the 2008 server it takes a reasonable time of 9 hours but the 2003 server is taking nearly 19 hours do backup the same amount of data.
When I look at the logs on the backup server to measure the throughput it is vastly different.
It is reported that the 2008 has a throughput of approx 51000.000 MB/Hour but the 2003 server only has a throughput of 23000.000 mb/hour.
I thought it could be something like the antivirus running at the same time as the backup. It was indeed running but since I have disabled this it hasn't made a difference in terms of the throughput.
I have read that it could be the network configuration on the VMware machines. The 2008 server is using the E1000 adapter which I hear provides much better performance than the flexible adapter. the 2003 has the flexible adapter,this server was originally done as a P2V migration and I think by default VMware gives the flexible adapter as the most compatible adapter for a VM.
I was thinking of changing the adapter on this slow server from a flexible one to either the E1000 or one of the VMXNet adapter. Are there any thoughts on this. I hear if i use the VMXNET3 adapter it simulates a 10gbps adapter in the VM even though we may only have a 1GBPS adapter. Has anyone seen any issues with this? I hear that this adapter will provide the best throughput and is compatible with the 2003 R2 Enterprise server I am running.
Any suggestions would be greatly appreciated.