Goals for Adump 1.2

I intend to make Adump 1.2 the first release of Adump that is actually useful for ordinary system backups. I have been using the 1.1 code (unreleased) to do regular backups, and am fairly confident of its robustness. Of course, other testers to find problems will be greatly appreciated.

Goals

The following are the goals of the 1.2 release.

Features now working:

• Full and incremental backups.
• Support for multiple dumps to a single tape.
• Media manager to determine and verify which tape is inserted.
• A new archive format. This eliminates most of the annoyances with 1.0 and trying to figure out where to record which files and directories had been deleted by the incremental.
• A small disaster restore utility.
• Archive compression. Individual files can be compressed in the backup. Zlib handles uncompressible data much better than my tape drive does. I was getting about 20% expansion on some of my data.

The following is left to do for 1.2.

• Simple client server. The server performs the actual backup. The client simply makes requests telling it what to do. This mostly improves performace, since the tape doesn't have to be rewound to recheck the label.

  GNAT 3.14p has been released, which has Ada.Sockets, making this effort much easier. I will use a simple authentication scheme, similar to Cram-MD5. No encryption will be supported, since there is no data being transferred.

• A set of scripts to perform normal backup operations. My intent here is that once a tape is labelled, you can place it in the drive, run a single script and it will perform the appropriate next dump to that tape. It should be possible to put this script in cron, and as long as a tape is put in the drive, a backup will happen. These scripts should also print out nice logs that can be printed (or emailed offsite) to be used to more easily locate the proper files to restore.
• Some basic documentation, explaining how to setup Adump and use it for common backup purposes.
• Instructions on how to build a disaster recovery disk. I'm not going to explain how to build a boot floppy, or CD, but at least explain what is needed (arecover and mt).

Future goals

The following tasks are needed (hopefully 1.3), but aren't planned for 1.2.

• Multi-volume spanning.
• Networked backups. Allow data to be transferred between the client and the server. There are some different ideas here, as far as lightweight clients, or lightweight servers. Will discuss more later.
• Encrypted backup channels. Perhaps this can be satisfied by tunneling the socket used by Adump. I also don't know enough about export law to know what I can and cannot do. AWS (link?) has a binding to OpenSSL which might be useful.
• A better restore utility. More information on the restore, as well as options for selective restoration. I still want to keep the small C restore utility, since it should be easy to build if needed. It also demonstrates the archive format succinctly.

Even more future goals

The following tasks are goals for beyond 1.3.

• CDR/CDRW backups.
• Database for fast restore of certain files.
• ...

This page last modified 2002/02/28.