Re: tbmode/tbtape on Online 5

On Fri, 20 Oct 1995 00:13:55, ian.goddard@geo2.poptel.org.uk (Ian Goddard) wrote: >In article <45juq2$3sp@news.Belgium.EU.net> cevi0004@pophost.eunet.be (Meyers Alex) writes: >>From: cevi0004@pophost.eunet.be (Meyers Alex) >>Subject: tbmode/tbtape on Online 5 >>Date: Thu, 12 Oct 1995 19:53:25 GMT >>We need an "continuous" backup scheme that allows us to tell our users >>(in case of failure or disaster) up to what time the system can be >>restored >>So we came up with following scheme >>Use large logical log files ( a day's work ???) in combination with an >>hourly crontab job that has to do the following things >>1) tbmode -l (lower L) to switch to the next logical log >>2) tbtape -a to archive this last logical log to >> a) backup tape >> b) Unix file (on another disk) >>The following questions arise What we use is : 1. Use many small Logs (so during high activity they fill about every 5 minutes) 2. Use many logs, so they last for 2 days 3. start continuous log Backup to a link to /dev/rmt1.1 4. Take the chance and tell everybody they won't loose more than 2h of Work This has the advatage of: + using less tape space, since in your case you end up backing up 1 tape block for archiving empty logs (we use 256k) + sometimes doing a backup every 5 minutes + leaving an absolute amount of work to be redone in case of catastrophe (in our case ca 5 minutes hard work of all people involved) + When we do a restore we can tell the restore time from online.log's last "time Logical log ## comleted" statement disadvantage: - if only 1 Person works he might work hours without backup - we take chances with the 2h statement Probably the best thing to do would be: 1. use small logs 2. use continuous logbackup 3. stop it every hour to backup the current log 4. then restart 2. If you use disk be sure to have an operator to keep an eye on the backup to tape. Always on the safe side Andreas Zeugswetter Sparkassen Datendienst AG&CoKG AUSTRIA / Vienna