Questions about the backup data size

We recently implemented a Networker backup system with Data Domain DDBoost devices as the backend storage devices. We noticed that there are big discrepancies between the backup data sizes reported by Networker, including DPA, and the consumed data domain storage space sizes reported by the Data Domain device. Two individual backup workflows were set up to show the differences.

The first workflow is to back up a VM whose storage usage size is 1.82TB. The VM has been backed up before with other VMs, now it is backed up individually. Before start running the workflow, we record the Data Domain file system usage by using the DD OS command ‘filesystem show space’. There is no other Policy/Workflow/Action running during the test. After the VM backup, the Data Domain file system usage was checked again. The difference between the two ‘Used GiB’ for ‘/data: post-comp’ is considered as the storage space used to store the backup data.

After the backup job, in Networker Administration GUI: Monitoring, there is a message: “Backup has SUCCEEDED (Moved: 113.92 GiB in 23m39s, Overall: 82.20 MiB/s, R: 114.90 MiB/s, W: 146.67 MiB/s).” What does this “Moved: 113.92GiB” mean? Is it the backup data size?

From NMC : Reports / Policy Statistics / Action Details, we saw the action’s “Amount of Data (GB)” is 1,860, while the “Target Size (MB)” is 562. I think this target size, 562MB, is considered as the backup data size, and would be reported by DPA as well.

We ran also the Networker CLI command ‘mminfo –S –q “ssid=xxx” against the save set ID of this job. In the output, we had

*ss data domain dedup statistics: "v1:1603116617:1903288930480:581839769:575350196";

575350196 / 1024 / 1000= 561.8654 (MB).

It seems that Networker thinks the size of the backed up data is about 562MB, although it is not clear what the “Moved: 113.93GiB” means.

But on the Data Domain side, the storage used during the backup is 94.1GiB (171x)!

The second workflow is a full backup for a physical host. The backup data size reported by Networker and that by the Data Domain device are quite close, at least at the same magnitude of order.

Then we ran the first workflow again. Due to the deduplication, the backup data size reported by Networker became 119MB, while the data usage size reported by the Data Domain device became 0.3GiB (2.5x)!

Does anyone know why the backup data size reported by Networker could be so different than the size reported by the Data Domain device? Is this related to the fact that Networker/Data Domain converts automatically an incremental backup for a VM to a full backup (I thought that conversion doesn’t require additional storage space with the exception of some extra meta data space)? Initially we thought we could use the backup size reported by Networker/DPA as an indicator for daily capacity consumption on the backend Data Domain devices. But with 10x – 100x differences, that backup size is useless!