Extract from Admin Guide Page # 263
Netbackup Admin Guide for Windows
(Netbackup Admin Guide is attached above in my post)
Load balance
The Load balance option allows NetBackup to select storage units based on a capacity-managed approach. NetBackup avoids sending jobs to busy media servers.
The selection is based on the following factors:
1 The rank of the media server. NetBackup considers the number of processes that are running on each CPU along with the memory thresholds on each server to determine the rank of a media server. If the free memory drops below a determined threshold, or if the number of running processes per CPU rises above a determined threshold, then the overall rank of the media server drops.
2 The number of jobs on the media server. NetBackup considers the number of scheduled jobs on each media server.
3 Whether the media server has enough disk space to accommodate the estimated size of the image NetBackup estimates the size of any new or current jobs on each media server and determines whether the jobs will fit on a given volume. NetBackup estimates the amount of space that the job may require, based on previous backup history. If no history is available, the high water mark for the storage unit serves as a guide.
(When is the Load balance option available?
Load balance cannot be selected for a storage unit group that includes a BasicDisk storage unit. Also, a BasicDisk storage unit cannot be included in an existing storage unit group with Load balance enabled. )
Other methods to distribute the backup workload
The Load balance option is based on the media server load and requires a license. The following list are methods to distribute the backup workload based on settings that do not require additional licenses:
■ To adjust the backup load on a media server:
■ Change the Limit jobs per policy policy attribute for one or more of the policies that are sent to a media server. To decrease Limit jobs per policy reduces the workload on a media server on a specific network segment.
■ Reconfigure policies or schedules to use storage units on other media servers. See "Limit jobs per policy" on page 103.
■ Consider changing the Bandwidth host properties on one or more clients.
■ To distribute the backup load on media servers during peak periods: Reconfigure policy schedules so that they write to storage units on the media servers that can handle the load. (This assumes that master servers and media servers are on separate hosts.)
■ To adjust the backup load on client Change the Maximum jobs per client global attribute. For example, to increase Maximum jobs per client increases the number of concurrent jobs that any one client can process and therefore increases the load. For more information, see "Maximum jobs per client" on page 432.
■ To reduce the time to back up clients Increase the number of jobs that clients can perform concurrently, or use multiplexing. Another possibility is to increase the number of jobs that the media server can perform concurrently for the policies that back up the clients.
■ To give preference to a policy Increase the Limit jobs per policy attribute for the preferred policy relative to other policies. Or, increase the priority for the policy. For more information, see "Limit jobs per policy" on page 103.
■ To adjust the load between fast and slow networks Increase the Limit jobs per policy and Maximum jobs per client for policies and clients in a faster network. Decrease these numbers for slower networks. Another solution is to use NetBackup’s bandwidth limiting. For more information, see "Limit jobs per policy" on page 103 and "Maximum jobs per client" on page 432.
■ To maximize the use of devices Use multiplexing. Allow as many concurrent jobs per storage unit, policy, and client as possible without causing server, client, or network performance problems.
■ To prevent backups from monopolizing tape devices
■ Place some drives in a down state or limit the number that are used concurrently in a specific storage unit. For example, if there are four drives in a robot, allow only two to be used concurrently.