Re: REPOST: RMAN question

Daniel Morgan <damorgan_at_x.washington.edu> wrote in message news:<1094102583.338898_at_yasure>...
> Howard J. Rogers wrote:
>
> >>What you describe could have, as easily, happened with a tape.
> >
> >
> > Of course it could. But how many times have you ever wanted to install a
> > spare tape into a user's PC because their tape had failed? So whilst it
> > could have happened to a tape, it is vanishingly unlikely that the
> > motivation ever to do it would arise in the first place.
>
> Never. But then if the tape had been properly stored at an off-site
> repsitory the issue wouldn't have come up. Same goes for a hard disk.
> And when it comes to backing up Oracle servers I try to keep those
> important backup items in the server room not in file cabinets.

Of course, how often do you validate the repository itself, aside from having a tape magically appear on order? Might be some guy with a U-Store just telling you about the gigantic underground caves. Or the caves might exist... and be next to radium deposits...

>
> Perhaps just another one of those American quirks. ;-)
>
> >>True. But hard disks are blazingly fast compared with tape and no more
> >>expensive per GB.
> >
> > Likewise true. But I thought we were talking about archiving and long-term
> > data storage? Not speed of taking a backup.
>
> We were. And I guess if the point is to have storage that will last for
> 50 years I couldn't hazard a guess as to which would be of value given
> that the operating system and hardware used to create it would no longer
> exist. But for any reasonable period I think hard disks are of equal or
> lesser price, just as likely to survive in proper storage and a heck of
> a lot faster.

Have you never heard the Banshee wail of a disk bearing entering lava lakes in Hell?

>
> > If you want speed, disks are good. Of course. They are always my first
> > backup device whenever possible, precisely for that reason.
>
> A point of agreement.
>
> > But if you want robust, long-term storage, then you forget the speed issue
> > and concentrate on robustness... which you've just agreed disks lack in
> > comparison to tape or other storage media.
>
> I don't recall agreeing on that. I recall saying I don't think either of
> us has ever had reason to pick up a 5 or 10 year old tape and find out.
> And I thought you'd agreed with that.

Remember Hollerith? He solved the problem of taking more than 10 years to do a count every 10 years. There are certain apps that need long term storage - government, insurance, health, science. The usual solution is to propagate data to new formats. The fact that that is often done in a destructive, no-rollback fashion is no reason to believe it should be done that way. That is just bad archival practice.

>
> >>I know a lot of people think that. And compared with diskettes, CDs and
> >>DVDs I'd agree. But I don't think the needs of an IT department are
> >>truly one of needing to restore to five years ago even though I tossed
> >>you that red herring. I think for the period of time when a backup might
> >>reasonably be used (1 day to 1 year) hard disks are as safe, if not
> >>safer, definitely faster, and in my opinion less expensive.
> >
> > Then we are talking about different matters, and not -incidentally- about
> > what the original poster asked, nor the point I was addressing when I
> > replied to him (and which I thought you were discussing).
>
> Probably true. I never get to these posts until close to midnight and
> after a bit of velvet smooth single malt scotch.
>
> > To remind you. The OP wrote: "I need to retain a few backupsets forever".
> > And I wrote "[backups are] a lot more safe on tape 'forever' than they would
> > be on disk"
>
> Ok "forever" but with the hardware in the dumpster years before. The
> operating system long since forgotten. Even if the tape survived ... so
> what? And I am still not convinced the hard disk wouldn't survive just
> as long.
>
> Someone saying "forever" is like someone saying "real time". If real
> time means instantaneously they are going to have to wait for quantum
> computing or the repeal of the laws of special relativity.
>
> > Now, you can re-define "forever" to mean "1 day to 1 year" if you wish, but
> > I'll decline that opportunity, if it's OK.
>
> Ok with me. Going back to my Cobol days and thinking about the year 2000
> I considered forever to be about 20 years and so did all of my
> contemporaries. ;-)
>
> My personal experience has been,
> > several times, a requirement to produce an archive that will survive for a
> > minimum of seven years (thanks to the tax and corporation laws here).
> > That's the sort of thing I think our OP was talking about, and what I was
> > addressing, too.
>
> Seven years in the U.S. for tax purposes too. And I've yet to find a
> 7 year old hard disk that failed sitting in secure storage: Perhaps
> others have.

See /. comments.

My PDP hard drive is 20 years old. I last tried it 9 years ago when I moved. Not secure storage, but my basement is actually a benign environment, being partly underground, on a gentle slope, and with HVAC because one of these days I'm going to convert it to my office and give my current office to my wife. One of these days, I'll find out if the non-pc will fire up and RS-232 from my linux box. The first thing after that is test the tapes. I expect the cheapo tape bearings will last better than the sophisto hd bearings. But if anything fails, I'm SOL, although I do have a second one rescued from a close call with dumpster.

>
> >>Well based on their confidence in Win95 and SCO one couldn't one draw a
> >>reasonable conclusion about the decision makers eh?
> >
> > You can sometimes be predictable! I knew you would want to seize on that
> > detail, so I mentioned it explicitly to try and head you off at the pass!
>
> I try to never pass up a chance for humor.
>
> > Ultimately, it makes no difference: we're discussing the long-term
> > viability of the storage media themselves, not the technology needed to
> > read those media. The O/S being used at the time is not a significant
> > modifier of the conclusions regarding the robustness of a particular
> > physical media format.
> >
> > These guys work for the British Museum. They do long-term storage (for
> > centuries) for a living. I think the results of their deliberations have
> > merit.
> >
> > Regards
> > HJR
>
> I agree with your assumptions about those folks at the museum. But I
> question whether anyone has actually run tests that bear this out as
> factual rather than anecdotal.

See stuff about rescuing old satellite tapes, and links I posted. It is a science.

jg

--
@home.com is bogus.
Two heartbeats from the presidency:
http://slate.msn.com/id/2106077/fr/rss/