Re: Upgrade to 8.1.7 or 9i

"Howard J. Rogers" <dba_at_hjrdba.com> wrote in message news:<3c1077e5$0$19079$afc38c87_at_news.optusnet.com.au>...
> Comments as ever...
> HJR
>
> --
> Resources for Oracle: http://www.hjrdba.com
> ===============================
>
>
> "Joel Garry" <joel-garry_at_home.com> wrote in message
> news:91884734.0112061720.6c6a2aa7_at_posting.google.com...
> > "Howard J. Rogers" <dba_at_hjrdba.com> wrote in message
 news:<3c0fcd82$0$29048$afc38c87_at_news.optusnet.com.au>...
> >
> > Since the main criteria is stability, I have to disagree.
>
> Feel free!
>
> >There will
> > be more bugs and product changes in 9i, simply because it is early on
> > the curve.
>
> That's where I think you're wrong. It's like saying that WinXP is worse than
> 95, because 95's been out longer and XP's a new release. It's the most
> stable first release of any Oracle version I've seen.

That really isn't saying much. The first release of any Oracle version is awful, and much stays that way for a long time. Having worked on 3, 5, 6, 7, 8, and a bit of 9 (and none of those on a first release) I can say there has been some improvement, but that still is a long way to go for someone whos _primary_ goal is stability. I'm not saying 9i is worse than 8i, I'm saying it is early on it's development curve. Well, actually I _am_ saying it is, _given the constraints_. You can't disable business constraints. For a commercial environment, you need to have some sound business reason to be on the bleeding edge.
>
> >8i has plenty of features for largeness,
>
> And missing ones which are crucial for true largeness. Materialized Views
> are good for 'largeness' -tried persuading the 8i optimizer to use them?
> Much, much easier in 9i. A whole raft of 'largeness; features that were
> *introduced* in 8i actually start working properly in 9i.

OK, let's go back and analyse what I was reading into the initial post. Some place not on the last 7 probably has some reason for that - perhaps they ran into some real problems with their app software, perhaps they are on some bizarro platform, perhaps they've been in lawsuits trying to make vendor offerrings work, perhaps their management has strange expectations of how much work it takes to keep a database going or upgrade it, perhaps all the original staff has left and this poor guy has to reverse engineer everything. Heck, a .de site might have a completely different idea of stability from language alone. And we really don't have an idea what large is - to me, a TB is starting to get there. I suspect the original poster thinks big is something smaller - the range of possibilities is large.
>
> >it's only real
> > problem is all the java stuff... which of course is the main reason
> > for 9, eh?
>
> Don't know what that means, really. 9i's main raison d'etre is RAC and all
> the other enhancements that make it a truly viable proposition for data
> warehousing, as well as the tweaks and twists which enhance its role as an
> OLTP solution.

We don't know what he's doing, but we can guess it isn't DW (although the possibility exists that it is a DW that was just stuck in a dev mode for a very long time - I've seen _that_ happen). I think you are overestimating the OLTP benefits - with an old app, it is entirely possible that it is still rule-based. Perhaps I am biased about that, as many enterprise and vertical apps I have seen still are, and yes they are current products. That happens a lot more than many people are willing to admit with mature products.

>
> >
> > If stability is truly the main criteria, I would go to 8i and watch 9
> > for at least a year. If you are on 7.3, you probably aren't even
> > using anything that would necessitate 9, but perhaps could use 8's
> > partitioning.
>
> And if he could use partitioning, perhaps he could use List Partitioning
> (Oracle 9i)? I think this is all the wrong way round. Of course, if he
> does a "straight" migration, and uses just the features of Oracle 7.3.4, he
> wouldn't need to upgrade at all! The point is, he probably wants to migrate
> in order to take advantage of new technologies... and 9i has more of them
> (and more useful ones at that) than 8i -Flashback, anyone?

I would venture a guess that that is a main reason he is still on 7.3.4. I would guess he doesn't want to migrate to take advantage of new technologies immediately - I would venture a guess it is a support issue. Maybe his boss got a sunset notice from Compaq :-O And perhaps a compatability issue if he has some old development tools trying to coexist with new.

>
> >If you want something like portal, you probably want to
> > put it on another box anyways. Even with 8, you'll likely have plenty
> > to do like getting rid of svrmgrl scripts and rewriting backup
> > procedures and all your other scripts that do anything useful. You
> > perhaps do everything with scripts? Do you want to replace all that
> > with manual operations? Even the new init.ora is now cat-juggling.
> >
>
> What does that mean?? There's not the slightest problem with the spfile as
> far as I can tell. It's a very handy feature.

Ok, that was a bad example, I thought I could get away getting some lurker nods from people that have a problem with it. But you didn't address the other issues in that paragraph. They are very significant in a commercial environment.

>
> Still, I'm not in the business of plugging the product. It's just that if
> you could have actually *listed* some instabilities in 9i, we might be
> getting somewhere. Just saying it's likely to have some because it's new is
> a bit of a non sequiteur and is, in this case, I think, demonstrably untrue.
>

The point is, I don't have to list them. It is not a non sequitur, it happens with every release of every product. Read the readme's from every O patch set... They break stuff! And see the other posts in this thread.

> Regards
> HJR
>
>
> > > 9i. It's robust, has lots of nice new features, can be migrated
 directly to
> > > from 7.3.4 (despite what the Oracle documentation itself says), and in
 many
> > > crucial areas is much more scalable than 8i (I'm thinking of dedicated
> > > agents for external procedures, multi-threaded heterogeneous service
 agents,
> > > and so on).
> > >
> > > 8.1.7 is de-supported in 2 years' time. Might as well hitch a ride on
 the
> > > newest version of the product, other things being equal.
> > >
> > > Regards
> > > HJR
> > > --
> > > Resources for Oracle: http://www.hjrdba.com
> > > ===============================
> > >
> > >
> > > "Thomas" <thomas.hiller_at_warema.de> wrote in message
> > > news:6e366956.0112060715.62df1b88_at_posting.google.com...
> > > > Hello,
> > > >
> > > > i have a more basic question. We currently run an very old Oracle
> > > > Server Version (7.3.4.1). Now we plan to migrate to a more up to date
> > > > version. Therefore my question is what Oracle Version to choose (8.1.7
> > > > or 9i). What are the advantages or disadvantages of the particular
> > > > versions. Let me know your experince about that.
> > > >
> > > > The main criteria is stability, the second is to improve the handling
> > > > of very big tables, the third the scalability - because it is
> > > > preditctable that the database is growing rapidly in the future
> > > >
> > > > What would you advice me ?
> > > >
> > > > Regards
> > > > Thomas
> >
> >
> > jg
> > --
> > Contrarian, as usual.
> > http://www.garry.to

jg

--
Glad to see people think, even if I'm right and they are wrong. :-)