Creating Jobs for Parallel Processing Solutions on a Single Subsystem
You can also create EVM jobs as part of a parallel processing system. For example, assume that you are implementing EVM to automate backup procedures at a site that has moderate resources. You have one StorageWorks subsystem with a moderate amount of free space and there are two tape backup units.
One solution would be to create several EVM jobs and the corresponding tape backup jobs. By designing script files or using a scheduler, you can have each EVM job replicate storage units and then start its associated tape backup job (see illustration below). As soon as the first EVM job has completed its storage replication, the next EVM job can be started, and so on.
Another approach is to create one EVM job that replicates all of the storage units that are to be backed up. For example, you might replicate units and mount them on the backup server as mount points /bcv_1, /bcv_2, /bcv_3 and /bcv_4, /bcv_5, /bcv_6. As soon as the EVM job has finished, you have the script file or scheduler start Backup Job 1 for mount points /bcv_1, /bcv_2, /bcv_3 and Backup Job 2 for mount points /bcv_4, /bcv_5, /bcv_6. When both backup jobs have completed, the script file or scheduler runs the undo-job to free up the storage assets.
When the snapshot replication method is available, parallel backup jobs will take only as long as the longest tape backup requires. No additional time is required to copy the source units.