[next] [previous] [contents]

  13.5 Designing Your Batch Queue Environment
  Use the following sample configurations to design your batch
  processing environment. Your configuration may combine
  elements from several of these examples.


  Configuration

                                                                                For More
                                                                                Information



  A single queue for limited batch processing
Section 13.5.1
  Multiple queues for heavy batch processing, or customized
  queues for specialized batch processing

                                                                                Section 13.5.2
  A VMScluster environment Section 13.5.3
  Queues that must be highly available Section 13.4



  This section provides examples of some typical batch con-
  figurations. Use these sample configurations to design your
  batch environment. Your configuration will probably combine
  elements from several of these examples.

  13.5.1 Simple Batch Queue Configuration: For Limited
              Batch Needs

  
Figure 13-1 illustrates a single, default batch queue. This
  simple configuration is suitable for a standalone system
  supporting mainly interactive processing.

  By default, when a user submits a batch job with the
  SUBMIT command, the job is placed in the queue named
  SYS$BATCH. To set up a single default queue on a standalone
  system, name the queue SYS$BATCH.

  13.5.2 Specialized Batch Queues: Meeting Special Needs
  If your users rely heavily on batch processing or have special
  processing needs, you might want to set up more than one
  queue. You can customize batch queues to handle specialized
  jobs by specifying performance and resource options for jobs
  in the queue. For more information, see
Section 13.8.5.

  The configuration illustrated in Figure 13-2 has several
  queues, each customized to process certain types of batch
  jobs.

  In this example, SYS$BATCH is the default queue. Normal
  batch jobs of no special importance would typically be submit-
  ted to this queue. FAST is a queue for executing high-priority
  jobs that should not be swapped out of memory. SLOW is a
  background queue for processing low-priority jobs. Typically,
  these are large jobs with large requirements for physical
  memory. Be conservative when changing base priority and
  swapping on a queue. Even a slight change can have a sig-
  nificant negative effect on batch and interactive performance.
  For example, even an increase of 1 in a queue's base priority
  can affect performance significantly.

  For information about specifying these options for a batch
  queue, see
Section 13.8.5.

  13.5.3 Generic Batch Queues in a VMScluster:
              Distributing the Work Load

  Generic queues can be used in a VMScluster environment
  to balance the use of processing resources by distributing
  batch processing across nodes in the VMScluster. (For an ex-
  planation of generic queues, see
Section 13.3.) Figure 13-3
  illustrates a typical configuration.

  In this example, a generic batch queue named SYS$BATCH
  is set up to feed jobs to execution queues on each node in the
  VMScluster. When a user submits a job to the clusterwide
  generic queue SYS$BATCH, the job is placed in the appro-
  priate execution queue to minimize the ratio of executing jobs
  to job limits for all execution queues fed by SYS$BATCH. For
  example, suppose execution queues MOE_BATCH, LARRY_
  BATCH, and CURLY_BATCH all have a job limit of 5. If
  MOE_BATCH and LARRY_BATCH are executing four jobs
  and CURLY_BATCH is executing one job, the generic queue
  SYS$BATCH will feed the next job to CURLY_BATCH.

  See VMScluster Systems for OpenVMS for more informa-
  tion about VMScluster queue configurations. For information
  about how to create a generic queue, see
Section 13.7.2.3.