Enterprise Volume Manager (Multi-OS) Version 2.0D Network Administration Guide

9–8 Enterprise Volume Manager (Multi-OS) Network Administration Guide
EVM Jobs and Job Management
The following example illustrates the use of a job built on a volume group or Tru64
UNIX domain without a specific fileset or mount point:
CLONE VOLUME dbserver VG(oracle_domain) 2
NORMALIZE VOLUME dbserver VG(oracle_domain)
SUSPEND WAIT dbserver freeze_db.sh
SPLIT VOLUME dbserver VG(oracle_domain) $BCV1
RESUME NOWAIT dbserver thaw_db.sh
MOUNT VOLUME $BCV1 backupserver 1 N/A
LAUNCH WAIT backupserver backup_db.sh
Using $<name> Variables
The split, snap, and mount operations require a $<name> variable or a unit number to
be specified. In nearly all cases, specify a $<name> variable and let EVM choose the
unit number automatically.
IMPORTANT: $<name> variables must begin with a dollar sign and may not contain spaces, as
in $copy_db1.
Specifying a unit number directly requires knowledge of the following:
• Unit numbers already in use
• Unit offset values for the EVM host agent where the BCVs are to be mounted
• LUN capacity of the EVM host agent
• Unit number range rules for snapshots
Running Multiple Jobs Simultaneously
Although the EVM job engine is multi-threaded and can handle the creation and
execution of multiple jobs simultaneously, some operations must be performed
serially, such as storage and EVM host agent access.
Because of these serial operations, there is a practical limit on the number of jobs that
can be executed simultaneously. In typical environments, four simultaneous jobs can
run with nominal performance degradation. As more jobs are submitted for execution,
however, the EVM GUI becomes less responsive to new requests and performance
starts to degrade.
The responsiveness of the EVM GUI to all requests (job creation and execution, as
well as resource and log requests) is inversely proportional to the number of jobs that
are currently running. Performance decreases as the number of running jobs increases.
NOTE: For optimum performance, carry out maintenance tasks, such as job creation and
editing, during periods when jobs are not running.