Are you Running Out of Backup Window?

The Problem

One of our customers opened a ticket recently with the issue that their production backups on their IBM i had to be cancelled because they were overflowing into their online processing the next day. Their daily backups start at around 2:50 am and were finishing around 9:50 am. They had been running the same backups of the same libraries and IFS objects for years, and they knew their backups were taking a little longer, but this was the first time they had to be cancelled. They asked Mid-Range to provide some advice.

The Options

Here are some things that were considered:

Moving some or all of the backups to the HA box to reduce or eliminate downtime for users on production.
Removing saves of save files from the backup
Upgrading to faster tape drives
Running saves in parallel
Converting to a virtual tape library (VTL) to provide encryption, an air gap, and faster save and restore times

The Solution

Analyzed their backups from Backup and Recovery Media Services (BRMS) logs
Determined which libraries and IFS directories were taking the longest
Discovered that many of the libraries that they were backing up were also being replicated to their DR box by MIMIX.
Discovered libraries were being saved to save files before production, after production, and then the save files were backed up too
Ruled out faster tape drives (too expensive), VTL (need an immediate solution), and parallel saves (not enough physical drives)
Decided on backing up from HA as the first step
The customer application team analyzed what libraries could be safely backed up from the HA system.
Business owner approved the changes based on the fact that the MIMIX product was running with almost no latency and no errors impacting libraries to be removed

After implementing the changes to BRMS and documenting the new restore procedures from the HA box, their nightly backup finished at 7:00 am (well before online processing was to begin)!