Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Extracting Redo Log Info/LogMiner/Shareplex

Re: Extracting Redo Log Info/LogMiner/Shareplex

From: Eric <bytewarrior2000_at_yahoo.com>
Date: 26 Sep 2001 12:14:31 -0700
Message-ID: <497b38b.0109261114.60bfa81b@posting.google.com>


Thanks for the info, Pete. I've got a few follow up questions.

We actually need the original rowid for a couple of reasons. We'll be going to 9i in some places, stuck indefinitely in Oracle 8i for other applications. Therefore we can't rely on supplemental log information allowed by 9i.

Also, once we have rowid, we use it to look up other information in the row which may not have changed in the update (e.g. we still need to know a customer's account number even if only his balance changed in a particular transaction). Obviously, if the row is migrated, logminer won't let us do that currently. Perhaps there's a way to use the current location of the row (physical address) in a query (select * from table where PHYSROWID = x?).

The other reason is that there may not be a primary key defined on the table, and no way to get a particular row without knowing the rowid.

Wondering how Oracle 9i Data Guard deals with this if it's doing statement-level replication on these types of tables? Or is there a restriction which mandates "insert only" unless there is a unique key to use???

Regarding LogMiner, START_SCN is allowed as you state. However, this appears to be a filter against the entire log, rather than a way to position intelligently into the log. In other words, the input is the same, but the output is restricted to the specified range. The problem therefore remains that we have to perform tons of IO in order to get the last few records.

Eric Received on Wed Sep 26 2001 - 14:14:31 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US