Re: Non-*.ctl Control File Naming Convention

On Jul 15, 7:57 am, Mark D Powell <Mark.Pow..._at_eds.com> wrote:
> On Jul 15, 9:44 am, "Joey.Dant..._at_gmail.com" <Joey.Dant..._at_gmail.com>
> wrote:
>
>
>
>
>
> > To all,
>
> > I had a vendor tell me yesterday, that it was a good idea to name all
> > of your database related files *.dbf, including control files. This
> > immediately set off a red flag as it violates OFA. Anyone doing this,
> > and if so what is your rationale?
>
> > The vendor said it made UNIX find commands easier.?!?
>
> > The vendor is a "reputable reinstaller of Oracle", but there were a
> > few other comments that made me question there knowledge.
>
> > A bit of background is that I just took a new role, and Oracle is
> > fairly new to the shop, and they have a large implementation. So they
> > had contracted the vendor to do the groundwork, before they hired me.
> > At this point it's a bit of an annoyance, as I'd like to just do my
> > own thing (well OFAs own thing) architecture wise, and not have to
> > deal with the 3rd party.
>
> > Thanks
>
> > jd
>
> And what exactly is so good about OFA?
>
> The software installation file structure and the database datafile
> file structure should be separate and distinct from each other.
>
> In the days of manual backup scripts naming all the database datafiles
> with a .dbf ending probably made some sense; but if you use rman for
> your backups when do ever have to script the database datafiles
> anyway?  You can query your datafile, control file, and log files
> names from the database so having .ctl, .dbf, and .log extensions
> makes sense to me since even if you do need to script them obtaining
> the names dynamically is not difficult.
>
> It could be a vendor support issue.  By using one standard then
> scripts probided by the vendor will work without changes being
> required due to difference in how sites name their files.  In my
> opinion the correct solution to this problem for the vendor is to
> write scripts that generate the scripts to do such tasks as create a
> manual hot backup, copy files to duplicate the database, etc ....
>
> HTH -- Mark D Powell --

On some systems, it can be dangerous to name redo files with a .log extension - you just never know when some helpful sysadmin finds and cleans up "useless old log files."

Of course, I don't always practice what I preach :-) but I notice I've somehow wound up with .rdo files on my 10G systems.

I think controlfiles, redo logs, archived redologs and database files are sufficiently different in function, especially in restoration, to have different extensions. I think the point about a standard is a good one, and an argument for OFA, as well as not changing from Oracle defaults without a reason.

It does seem to me (consistently over years of observation), that many vendors just aren't making the effort to research the issues and do things right from the Oracle point of view. They just pay money to be an Oracle partner so they can sell more services.

In the end, these are arbitrary DBA decisions, which the DBA and everyone else in the future will have to live with whether or not he is the one who makes them.

jg

--
@home.com is bogus.
http://www.signonsandiego.com/uniontrib/20080715/news_1b15indymac.html