The Makers of FDR®

FDRDRP Technical Description

In this section we expand on the example that was introduced in the overview section.

Although this is still a highly simplified example, it perfectly illustrates how FDRDRP operates and where it achieves the impressive savings on DR times.

The diagram shows the backups of three DASD volumes--PROD01, PROD02 and PROD03. Full-volume backups of these disks were taken at the weekend with FDRINC, and written to tape 111111.

Daily incremental backups were then taken on Monday and Tuesday evenings, written to tapes 222222 and 333333 respectively.

Ordinarily, the standard FDRINC Full-volume reconstruct process would have to mount and rewind each of the tapes 3 times (9 mounts/rewinds) and would take time to position to the required backup file.

FDRDRP, on the other hand, would mount and rewind each tape only once (3 mounts/rewinds) and it would eliminate all of the positioning delays. This results in a typical elapsed time saving of over 80%.

The Process
FDRDRP initiates a recovery subtask for each of the three DASD Volumes, which are then sorted by the tape volser and fileseq required for the first backup. This allows the subtasks to read the backup files on a tape in physical order with minimal positioning.

At the start of a recovery process, FDRDRP would initiate a subtask for each of the 3 DASD volumes.

The subtask for PROD01 would dynamically allocate and mount cartridge 333333 and then start the restore from File 1 (PROD01’s Tuesday incremental).

The subtasks for PROD02 and PROD03 would wait for cartridge 333333.

When the PROD01 subtask finishes with cartridge 333333, the restore subtask for PROD02 would pick it up (without rewinding or dismounting it) and begin the restore from File 2, which is PROD02's Tuesday incremental.

At the same time, the restore subtask for PROD01 would then mount cartridge 222222 and start the restore from the Monday incremental

PROD02’s subtask would continue to restore from cartridge 333333 and then hand it over to PROD03’s restore subtask, in turn picking up cartridge 22222 once PROD01’s subtask had finished with it.

And so the process would continue until all the backup files on the three cartridges had been restored.

No Changes Required To Backup JCL
All of the intelligence of FDRDRP is implemented within the restore phase. It is not necessary to alter the FDRINC backup JCL in any way. This means that you can achieve the impressive savings in DR recovery times, without having to alter the backup processes you already have in place.

Restore Example
Here is a simplified example to illustrate using FDRDRP to concurrently restore 10 DASD volumes from their FDRINC backups.

//SYSIN       DD  *


  • The 10 volumes indicated on the SELECT statements will be restored in parallel.

  • Since the volumes at the disaster site are pre-initialized to a known volser, each SELECT specifies VOL= for the volume to be restored and NVOL= for the target volume to restore to. Each target volume will be relabeled to the serial of the restored disk.

  • FDRDRP will locate the most-recently created COPY 2 backups for each disk volume in the ABR catalog, which must have been previously restored.

  • No more than three tape drives will be used by these restores. FDRDRP will mount the backup tapes the minimum number of times necessary to restore data for all disks. All required disks and tapes will be dynamically allocated.

return to top