Brian Aker (of MySQL) graciously stepped in to answer some questions and provided reasons for some of the design decisions in MySQL OnlineBackup. Some of them are:
- Envelope format for backup archive. This would allow backup as well as recovery to continue in case the storage engines failed to do backup or recovery respectively.
- Block size and envelope format will help in handling tape errors.
- Logical backup will be done in case storage engine does not have online backup API implementation.
- Status of Backup run (BACKUP SQL command) will be available as part of MySQL information schema.So, backup can be run as a truly background process.
- Online Backup API will be implemented by all storage engines (except for NDB storage engine) in the first release.
- MySQL OnlineBackup will depend on external management tools such as ZRM for MySQL for backup scheduling, reporting, monitoring, compression, encryption and other enterprise features.
My presentation slides are here. ZRM for MySQL will implement MySQL Online Backup as one of the backup methods. For people who could not attend the session, I will be hosting a webinar on Nov 16th 10am PT on the same topic. Please register for the webinar.