Restore Flow for Long-Term NAS Backup Archival

[dulgidulgi]

Recently, a customer reached out with an inquiry. They are planning to replicate their NAS backup data to the cloud, with an expected retention period of 10 years. They wanted to know how the restore process would work in a disaster scenario where a full restore of the NAS is required.

As far as I know, the NAS backup workflow performs an initial full backup followed by forever incremental backups.

My question is about how a full restore would behave after 10 years of data retention if we have to restore solely using the data stored in the cloud. Does the system need to reference all backup data across the entire timeline?

For example, suppose we started replicating NAS backups to the cloud in 2016. If a disaster occurs now in 2026 and we need to perform a full restore:

1. Does the restore process actually start from the 2016 full backup and apply all subsequent incrementals?

2. While I understand that NAS backups typically manage the latest active state of data using metadata, what happens to files that have remained unchanged since older periods (such as 2016 or 2017)? Does the restore engine have to scan and reference every single backup point across the 10-year span to reconstruct the volume, or does it directly identify and pull the required blocks/files?

I would appreciate a detailed explanation of the restore workflow under this scenario.

[david.domask]

Hi dulgidulgi,

First, give a quick look at the structure of an Unstructured Data Backup; it's not the same as what you're thinking with VBKs & VIBs, it's better to understand it similar to a database without a database engine.

Backup, archive and secondary repositories store file and object storage backups as objects in VBLOB files (up to 64 MB each) of unstructured data. They also store metadata about the original objects on unstructured data sources and their structure.

By using a model like this, the Unstructured Data Backup can better manage the file versions and only read / recover from blobs that contain the necessary data.

So to answer your questions:

1. Data from blobs containing the file versions applicable to the chosen restore point will be read from
2. No, the metadata allows the restore to understand exactly which blobs have the necessary data, and the Unstructured Data restore engine will read exactly what it needs to.