Lots of questions about backup and restore

I recently planned to design a collaborative backup restore system based on foundationdb and other systems. After investigating the documentation, forums and issues, I have many questions:

  1. backup and restore

    • What is the progress of the new backup system (v2)? I only see one parameter in the documentation
    • What is the progress of the new restore system (Performant restore)?
    • Will DR base on the new backup restore system in the future?
  2. DR

    • I read the forum that there is a pause and resume function, but I don’t see it in the documentation
    • I saw the issue pause and resume had performance problems, is it solved?
    • Abort current DR, then start again, can continue to synchronize based on current data, incrementally? Or is it equivalent to start from empty data?
    • Is there any way to achieve this efficiently?
  3. Hybird

    • File snapshot-based backup restore

      • Is there a detailed description of how to use it, including features and limitations?
      • Guess: first get the system into some kind of state and then the user goes to snapshot accordingly, or is everything organized by fdb?
      • Can snapshots be imported and exported?
    • Point-in-time restore with DR

      • What are the limitations? For example, restrictions on atomic operations
      • Do I need to pause or abort DR?
      • How to get the starting point of the restore operation? issue?
  4. multi-key range backup restore, suppose there are two key ranges: A, B

    • Is there a performance difference between a non-separate backup and a separate backup of A and B?
    • Does the range division of backup affect restore? For example, is it possible to do separate restore of A and B without separate backups?
    • Is there any performance difference between non-separate restore and separate A and B restore?
    • If the goal is to restore A and B separately, is it better to choose not separate backups or separate backups?
  5. Incremental restore

    • What is the progress of the incremental restore? Has it solved the problem of atomic operations?
    • Is the new backup restore system required?
    • Does it support continuous incremental restore? For example, the first time from version 1 to version 100, and the second time from version 100 to version 200?
    • Does it support incremental restore with multiple key ranges separated? For example, first restore range A from version 1 to version 100, and then restore range B from version 1 to version 100.
    • How can I pause an incremental restore and then continue? Will there be DR-like performance issues?
    • How to abort one incremental restore and then continue? Where can I find out how to continue (the begin_version to continue)? For example, can I check where the current restore has gone by stats?
    • If the backup is performed continuously, if the restore does not select the target version, will the restore be performed continuously? Or does it stop automatically? How do I know how to continue next time after stopping?
    • Is there any plan or possible support for version rollback?
  • What is the progress of the new backup system (v2 )? I only see one parameter in the documentation
  • What is the progress of the new restore system (Performant restore )?

The new backup is ready and included in 6.3.
The new restore is not ready, You can observe the progress here
The format of the new backup is not compatible with the old restore. Because the new restore is not ready, the only way of using the new packup is to convert backups to the old format with a special utility

The problem still exists in 6.3. It has been solved in 7.0

It is not possible. Starting replication Primary → DR requires the DR database to be empty.

Yes, it is. To use pause instead of abort.