Re: Best backup scheme of single tape drive system

In article <359A4012.6DA0@bloomberg.net>, kagel@bloomberg.net wrote: > > Chuanlarp Satchavarodom wrote: > > > Dear List, > > > I have a question about continuous database backup for a single tape > > system. I want to do continuous logical log back up, but my system has > > only one tape drive. If I dedicate it to back up full log file, when I > > want to do dbspace backup, I am gonna have a problem. =20 > > > This is what I plan to do, I will set a logical log backup to tape, > > and when the schedule time for dbspace backup arrives, I will stop the > > continuous log back up. Backup dbspace and perform automatic logical > > log backup and turn on the continuous log again. Does this make sense? > > Is there a way to better automate the whole thing with single tape = > > system? > > You can simulate continuous log backup to disk with better safety than > using ontape -c by using an ALARMPROGRAM script or executable to launch > ontape -a each time a logfile is completed. A sample for doing this> using onbar is included with the engine in $INFORMIXDIR/etc and I have > both a simple script and a "C" executable to do this using ontape -a. > I submitted the package to the IIUG Repository this week but it will be > a few weeks before it is posted on the web. Let me know via email if > you want a copy. The idea is to archive the logs to disk, rename the > file so that it is not overwritten by the next log backup, then later > copy the log backup files to tape with the normal system backup > schedule. Since the logs are primarily needed between Informix backups > you should schedule system backups between and immediately preceeding > the Informix backup and the Informix backup script can then delete any > log backup files older than the previous database backup on the > assumption that they have been backed up already and are no longer > needed. > > Art S. Kagel > Let me add a little detail What we do here is like this (We are on V7.12 - onbar is not available) We do a continuous Logical Log Backup to disk using onarchive. The disk we put the files on is mirrored by the OS. And we do DB Backups Level 0-1-2-1-2 every evening during the week (no 7x24, rather 5x14). The disk we put the archives on is not mirrored. In the nite we run ADSM to copy & delete all these files. What I had to learn the hard way was - at least using ondatartr for RESTORE - you have to have ALL the Logical Log Backup files since the most recent archive level 0 to be able to do the Logical Recovery (i.e. RollForward) after a Cold Restore. Processing is fairly quickly and it should IMHO only be necessary if you don't do archives level >0 of all the dbspaces together in one run as we do. (During processing of the Logical Log Backup files you have a checkpoint line every 5 second according to online.log - so it is used to keep the different archivings of the noncritical dbspaces in sync I suppose) BUT as I mentioned: You have to have ALL the Logical Logs since your last level 0 - no matter if you end your Cold Restore with level 0, 1 or 2, the Logical Recovery starts with the Logical Log File containig the checkpoint record written immediately after your last archive level 0. The only situation you don't need Logical Log files at all is when you switch to multiuser using onmode -m after the cold restore and do no Logical Restore at all. Then you are in the state you had at the time of this archive level x, which is the last one you feed in This is well tested on INFORMIX V7.12 UC1 on SNI SINIX V5.43 and is maybe only so with onarchive Good luck Dick ----- Richard Kofler EDVg-debis Systemhaus GmbH & CoKG Vienna / Austria -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum