<?xml version="1.0" ?>
Yes, auditdb binary output is piped into the Java program the builds the JDBC commands to execute them. A unique key is required.

-----Original Message-----
From: martin.bowes@ctsu.ox.ac.uk [mailto:martin.bowes@ctsu.ox.ac.uk]
Sent: Thursday, June 09, 2005 3:26 AM
To: Kale, Mark (STP)
Cc: info-ingres@cariboulake.com
Subject: Re: [Info-ingres] Re: Ingres stand-by database


Hi Mark et al - gee this is getting interesting,

> Jumping on on the thread late, but thought I would give a comment.
> Each night we do exactly what you are saying. We do a production
> checkpoint, transfer yesterday's journals and the config to the drp
> server, and do a rollforward. We follow with full checksums. It is
> not that hard to do.

Absolutely so. But this is not a creeping or rolling recovery. This sounds like a normal 'static' recovery to a foreign host.
>
> We also have a custom-built Java replicator that reads yesterday's
> journals from production, and applies certain table DML-only commands
> on our reporting server. Works great.

So you use your Java program to mimic auditdb audit trail or VDBA journal analyzer to produce sql commands such as insert, update, delete to work out the sequence of commands to apply to your recovered database to bring it to some point after the checkpoint?

This could work well - with a caveat on tables with no unique key. You may also need to be careful of floats and any byte data types.

Marty
--
Random Duckman Quote #46:
Duckman: Damn I must have missed one...Um, I mean cram, unjust, math, piston.
Cornfed: Uh huh. Are you having a stroke?
Duckman: I haven't decided yet.