Re: RDBMS Feature Matrix (FAQ in the works !!!)

In article <C4CLCs.6JM_at_unixhub.SLAC.Stanford.EDU>, ian_at_tethys.SLAC.Stanford.EDU (Ian A. MacGregor) writes:
|>Subject: Re: RDBMS Feature Matrix (FAQ in the works !!!)
|>From: ian_at_tethys.SLAC.Stanford.EDU (Ian A. MacGregor)
|>Date: Tue, 23 Mar 1993 15:22:02 GMT
|>
|>|> In article <1993Mar22.151030.18234_at_ryn.mro4.dec.com>,
|>|> abbott_at_priory.enet.dec.com (Robert Abbott) writes:
|>|>
|>|> This sounds like mirroring of the log at the file level rather
|>|> than the device level. Correct? There is still only one logical
|>|> log per database, although it may have multiple physical copies.
|>|>
|>|> My question was if Oracle can support two or more logfiles
|>|> per database where logfiles are not identical i.e., some logging
|>|> is directed towards one file the rest to another.
|>|>
|>|> ------------------------
|>|> Robert K. Abbott
|>|> abbott_at_tps.enet.dec.com
|>
|>Your analysis is correct. There is only one redo log group ONLINE at any time,
|>and yes using groups is akin to mirroring at the file level. As I stated in my
|>last post, you must have at least two groups (if you choose to use log file
|>groups) so that a log file switch can be performed. There is no requirement to
|>use groups in Oracle7. You can use the same log file setup used in Version 6.
|>
|>It sounds like you want two or more groups of log files to be active at any one
|>time. No, this is not currently possible. I cannot see how you would inform
|>the log writer process what information to write to one group and what information
|>to write to another.
|>
This is not EXACTLY true. When you run in parallel mode, each instance uses a different set of log files groups. Thus, each instance has a different active log file at any given point in time. This behavior is a change from parallel mode under V6. The V6 behavior was to have all instances share the same active redo log.

My guess is that Mr. Abbott really wants to know if you can split the logging for an instance rather than a database. (Which you can't do.) But you can have multiple instances running against the same database on a single machine. The reason normally given as to why someone would want to do that is that you can tune each instance for different "workload types". In this case your tuning would be to have different sets of redo log files (presumably on different devices) to spread the redo I/O between files.

Brett E. Bowman
Eli Lilly & Company
BBowman_at_Lilly.Com Received on Tue Mar 23 1993 - 21:16:01 CET