Avamar shows up to ~30% usage even after all backups have been deleted and garbage collected

An Avamar grid has been used for backing up data. 

The user carries out the following actions:

• All backups are deleted from the Avamar grid, including backups under the /REPLICATE domain.
• Garbage collection runs repeatedly (over several days). There is no more data left to be garbage collected.

After these actions, the "utilization" level drops, but does not fall all the way down to zero as might be expected.

The space consumption which the utilization value continues to report is space taken up by Avamar parity overhead.

Parity stripes provide redundancy across nodes and protect against single node failure.

Parity stripes consume space on Avamar. Unlike regular data stripes, once a parity stripe has been created, the Utilization calculation always considers it to be consuming space.

This is so even when a regular stripe contains no backup data.

A parity stripe would still calculate the parity of a stripe that is filled with zeroes.

The fewer nodes in Avamar, the larger the "utilization" level shows when this occurs.
 
This is similar to the efficiency of RAID 5 disk arrays where space equivalent to one disk is required for XOR parity information. The fewer nodes, the less space is available for the stripes of the parity group.

This is normal and expected behavior. No corrective action is required.

If the grid is once again filled with backup data, the primary stripes are populated and the utilization figure increases.

The rate of increase is slower than when the grid was originally filled with data. This is because the space consumption of the parity stripes is already factored in to the 'utilization' figure.

Parity stripes do not contribute to the increase in 'utilization' as new backups are added to the grid.