I have a dedicated media server attached to an LTO5 tape drive and autoloader that performs the vast majority of our file and folder level backups. Most of the machines that are being backed up are virtualized on VMware. We also have the VI agent for Backup Exec 2010 R3 and I have configured only two machines to be backed up using this agent. One is our AD controller and the other is our Exchange server. These backups occur on different nights and after the tape-based jobs have finished. The backups are going to a CIFS share on an HP D2D4004i backup appliance. The backup appliance is configure with 2 1Gb connections setup in a Link Aggregate configuration and the switch that they are connected is also configured with LACP. The media server has a NIC dedicated to the backup appliance's network. The AD controller VI job is configured to use GRT and the Exchange backup is not. The average throughput for those jobs is around 175Mb/s. The average for the tape-based backup jobs is anywhere from 1600Mb/s to 2600Mb/s (depending on the job type). Nothing else is backing up to these appliances at the time of the VI backup. I'm trying to isolate why the VI backup is taking so darned long. The problem I am running into is that it is not completing in time. I have two D2Ds and they are replication partners but the replication won't start until the job has ended; so I am pretty sure that there is no overlap or contention for resources there. I have checked various best practices and white papers and all seems to be configured properly. I am at a loss as to why the pathetic throughput. Any ideas?
↧