How to configure NetBackup Resource Broker (nbrb) to improve resource allocation performance

Description

In certain environments, the time it takes the NetBackup Resource Broker ( nbrb) to execute through its queue of jobs waiting for resources can limit overall system performance. This may occur in situations where many jobs are queued for resources and the jobs are completing faster than nbrb's ability to re-use the released resources for the queued jobs.

Beginning in the NetBackup 6.5.2 Release Update,the nbrb.conf can be utilized to improve the performance of nbrb. The nbrb.conf file is located in:

UNIX: /usr/openv/var/global/nbrb.conf
Windows: <install_path>\Veritas\NetBackup\var\global\nbrb.conf

This file offers the following parameters:

• SECONDS_FOR_EVAL_LOOP_RELEASE
• RESPECT_REQUEST_PRIORITY
• DO_INTERMITTENT_UNLOADS

Note: By default all values are 0.

Example format for parameters in nbrb.conf

SECONDS_FOR_EVAL_LOOP_RELEASE =180
RESPECT_REQUEST_PRIORITY =0
DO_INTERMITTENT_UNLOADS =1

Explanation of nbrb.conf parameters

• SECONDS_FOR_EVAL_LOOP_RELEASE
  Default value is 0 if not present in nbrb.conf.  If the value is 0, nbrb reverts to normal default behavior, evaluating all queued job requests before releasing any drives that have been released by completing jobs.
  
  When set to a nonzero value, SECONDS_FOR_EVAL_LOOP_RELEASE controls the time interval after which nbrb will break into its evaluation cycle and release drives that have been given up by completed jobs, making them available for use by otherjobs.
   

• RESPECT_REQUEST_PRIORITY
  Default value is 0 if not present in nbrb.conf; only has effect if SECONDS_FOR_EVAL_LOOP_RELEASE is set to a nonzero value.
  
  When set to a nonzero value, nbrb will restart its evaluation queue at the top of the prioritized job queue after resources have been released due to the SECONDS_FOR_EVAL_LOOP_RELEASE time interval having passed during the nbrb evaluation cycle.
  
  If set to 0, causes NBRB to continue evaluating jobs in the prioritized job queue at the point where the evaluation cycle was interrupted for drive releases due to SECONDS_FOR_EVAL_LOOP_RELEASE interval. This mode of operation may make it more likely for a job to reuse a drive more quickly after the drive has been released, at the expense of allowing some lower priority jobs to get drives before higher priority jobs.
   

• DO_INTERMITTENT_UNLOADS
  Default value is 0 if not present in nbrb.conf; only has effect if SECONDS_FOR_EVAL_LOOP_RELEASE is set to a nonzero value.
  
  When set to a nonzero value, nbrb will initiate unloads of drives that have exceeded the media unload delay when resources are released due to the SECONDS_FOR_EVAL_LOOP_RELEASE time interval having passed during the nbrb evaluation cycle. This will make drives available more quickly to jobs that require different media servers or different media than the job that last used the drive. This is achieved at the expense of not having the loaded media/drive pair available for jobs further down in the prioritized evaluation queue that could use the drive/media without unload.

• If nbrb.conf does not exist, or if parameters are not set in the file, then those parameters take on their default values (0), and nbrb will behave as it normally does without these configuration parameters.
• The addition or modification of the nbrb.conf file does not require a recycle of the NetBackup processes.  The process will read this file at the start of every evaluation cycle, and changes of any type will be implemented at that time.