Re: Any APIs for Backup and Recovery ?

"GMg" <godmanng_at_yahoo.com> wrote in message news:7b9aab6a.0406051613.747f7f29_at_posting.google.com...
> > There aren't any. Or do you want to create your own RMAN?
> > The RMAN procedures are available by means of the datadictionary.
> >
> > I'm not sure what you are up to. With the RMAN tool being available I
> > don't think there is any need to reinvent the wheel.
>
> Thanks for answering. I am not planning to reinvent any wheel.
> Wouldn't it be good if I can just write a simple java code to call
> those APIs (granted they are available) and do an online backup. Our
> size is about 120GB and I think I still can survive without using
> RMAN. I am using the conventional Alter tablespace begin and end
> backup. I don't see the advantages over using RMAN at my site at this
> moment. Maybe someone can convince me to use RMAN, but I doubt it.
>
> Anyone actually using RMAN and how do they like it compare to the old
> way? Pardon my ignorance, I am not in this business long enough to
> judge which is the way to go. However, from a developer background, it
> will be neat if those APIs are available.

I'm confused. Since you do things the 'begin backup...copy....end backup' way, what API's do you need? You simply need your Java code to issue plain SQL statements and muck around a bit at the file system level. At the www.dizwell.com Backup and Recovery FAQ, I have a WSH (Windows Scripting Host) script which "programatically" achieves the same thing. You might find it useful as an example of what you'd have to do using Java calls.

I won't bother trying to convince you of the merits of RMAN over the O/S way of doing things, but they are many, including inherent validation of the backup against corruption; no redo overhead when doing hot backups; ability to skip never-used blocks, and thus produce much smaller backups; ability to recover block-by-block instead of data file by data file (if you've got 9i) whilst nevertheless keeping the rest of the data file online and open for business; ability to incrementalise your backup strategy; much simpler incomplete recovery procedures; cloning capabilities; and so on.

It rather depends on your version, I suppose. If you've got 8.0, there are good reasons for sticking with the manual technique. But from 9i onwards, I seriously think anybody who develops new backup strategies that involve manual O/S-based activity, regardless of the size of their database, is (a) asking for trouble and (b) making work for themselves to no great benefit. If you see where Oracle is headed with 10g and so on, it is clear that manual backup/recovery methods have no real future, either.

Let me put it this way. The RMAN command in 9i is "backup database;". How much programming do you want to do in order to achieve that result?

Regards
HJR Received on Sat Jun 05 2004 - 19:34:17 CDT