Open firebird-automations opened 21 years ago
Commented by: Alice F. Bird (firebirds)
Date: 2006-01-23 21:17 Sender: kpalagin Logged In: YES user_id=838827
I absolutely agree. And this is not only Oracle feature, but MS's too - ESE has it. Redo log can provide significant performance improvement in disk access by allowing separation of sequential uncached writing of transaction log and random buffered writes and reads.
Commented by: ohenri (ohenri)
Every well-designed DBMS has this feature, except Firebird - but the lack of this logging makes a Firebird databse easy to handle.
So I would like to extend this request as follows:
- Choosed on a "per database" way, it should be possible to switch on/off a full logging system (in addition to the normal generation system).
- Logs shall be written to local drives
- Dual-Logging shall be possible too (where at least one log may be placed on a network drive), where logging continues if one log is lost (zero space conditon, for example).
- Partially failed dual-logs shall be merged by a tool to a healthy single log.
- Having a backup and all consecutive (single) logs (from that point), a roll forward shall be possibe.
- The roll forward shall be limited to (stopped at) any specified date/time .
Submitted by: mgast (mgast)
Votes: 1
SFID: 819330# Submitted By: mgast
Description of the current state:
1. In Firebird all data changes applied to a database after the time of starting the last database backup a lost in case of a failure that corrupts one of the database files.
2. If your last database backup is corrupt, you only can switch to the state stored in an older backup.
What i would like to have is a kind of REDO LOG facility as it exists in Oracle RDBMS. In Oracle it is realized as a buffer that works on a set of files (organized as flat sequence or as ring buffer) in which all statements changing the content of the database are stored.
In the case of a corrupt database file, you can rebuild a state of your database later than the point of time, where you started your last database backup.
If you backup all redo log files used since the starting time of a given backup you can rebuild your database to every state after this backup. A corrupt database backup does not longer result in uncontrolled data loss.