Oracle hype for new features

[guru_heaven]

Originally posted by jmodic
And BTW guru_haeven, Oracle *does* release BETA versions and some customers and partners indeed *are* their BETA testers.

Sure they do and there are *REAL* beta testers around, but we all here are their *FREE* beta testers and that's why they get bad reputation day after day. I have been working since Oracle 6 and until Oracle 7.3.4 thing were OK, after that they just release unfinished software to market to speedup time-to-market strategy.

We have 120 production instances and half of them is still running 7.3.4, it's not that we dont want to migrate it's that all applications we have most are in-house which have tons of PL/SQL, PRO*C codes, we asked Oracle for a migration plan to migrate to 8.1.7.4 or 9.0.2.1 and they dont want to get involved. Guess what because they cannot guarantee the PL/SQL and PRO*C codes can still work in new versions, funny huh.

OPS is another toy which never worked properly and you must adapt your application to it, we had 3 OPS instances and finally got rid oof them, instead of improving performance they degrade, OK you may say it's application's fault and that may be true but of course we are not gonna develop an massive billing application to adapt to a RDBMS vendor. Mr James Morle published an UNBREAKBLE RAC whitepaper recently and he showed that RAC is stilla lousy feature, Oracle says RAC is transparent however from James Morle's tests that's certainly not true. The waits, latch contentions are much much higer in RAC than a non-RAC environment, so can we say RAC is still a toy?

[jmodic]

Originally posted by guru_heaven
Sure they do and there are *REAL* beta testers around, but we all here are their *FREE* beta testers and that's why they get bad reputation day after day.

Can you give us one single example of large software package nowadays where things are not as you hacve described them? Oracle is not an exception here, not positive nor negative, IMHO.

I have been working since Oracle 6 and until Oracle 7.3.4 thing were OK, after that they just release unfinished software to market

That is another broad generalization that is simply not 100% true. When Oracle 7.0 came out it has become realy bad raputation because of numerous problems. Then again with 7.2 as far as I can remember. So no, things were not so great untill 7.3.4 either. But this is quite normal, I would say. Each release has many new features and each release fixes many known bugs, but it also potentialy brings many new bugs. This is simply inevitable.

But is this the reason to simply say: "all new features Oracings are dangerous to use in bussiness"? If this were true we would still be using Oracle6 in all its glory. We would not use some basic features that we now take for granted, like PL/SQL in a database for example. This also once was one of those "new and fancy features" of Oracle7.0 that was also quite buggy at the time of the release. Is farrokhp using PL/SQL in the database nowadays? Probably yes, although it was once one of those "new and fancy features" that drove him to the conclusion that "it is basically dangerous to use it in business".

[jmodic]

Originally posted by farrokhp
Hey jmodic, I don't know what background do you have but try to have ethic and be polite if you want to talk, obviously you have bad attitude.

Hej farrokhp, would you care to explain what drove you to that conclusion? I've read my post again and couldn't find a particular reason for the offence.

Maybe the fact that I didn't call you names, like "Mr. genious" (speaking of ethic, attitude and politeness....))? If so, I appologise and promise that I will never make that mistake again. Or is there any other title that yopu choose?

Maybe the folowing sentence has raised your pressure? "I would say your conclusion that "all these new and fancy Oracle features is basically dangerous to use in business" is bunch of crap." If so, I take it back and appologise sincirely.

However I can't help myself. Whenever I'll see a statement that "all these new and fancy Oracle features is basically dangerous to use in business" I'll hide in some lonely place and whisper to myself so that nobody could hear me: THIS IS NOTHING BUT A BUNCH OF CRAP!

[marist89]

Originally posted by farrokhp
They prefer to pay big bucks to veritas rather than deal with cumbersome fail over oracle offers.

Veritas has been doing clustering and failover for years. Oracle has been doing failover for months. You choose.

I knew that in Canda only 2 or 3 major shops use practically parallel servers. Goverment of BC tried to use it a while ago, hit lots of problems and lack of expertise then they switch back to EE servers.

I'm sure jmodic is thanking you for making his point. Just because they didn't have the experience doesn't mean the features are worthless.

If you have a mission critical production db is better play safe as much as you can. Even if you test, you can't still %100 sure. Can you? So Mr. genius jmodic
this is again bunch of crap... Haaa?

Of course you can't be 100% sure. However, with the proper amount of testing, precaution, and experience you can rely on the features provided to do the job they were intended to do.

[farrokhp]

I agree that I might express that a little too strong when I said " all new features are dangerous to use in business" , That was not the main intention. What I am trying to say here is from manager point of view , assuming that you are a top notch DBA and you figure out these new stuff in a glimpse of a time, but if you left company and they can't hire another genius Jmodic!!!, then they are in trouble. So 9 times of 10 they decide to stay with core features. Also what I am saying is new features takes a while ( usually long time ) to become real practical solid features, I don't want to be victims of their bugs in the first place, doesn't mean they are useless but for the time being they are more marketing triggers to create more sales then eventually they will become either obsolete or real features.
Ok Mr. Jmodic, you sounds very political to me, you made a fake apology but then you repeat your insult this time with uppercase letter : THIS IS NOTHING BUT A BUNCH OF CRAP! , thank you for your high level of ethics. Kind of person that politely rude and screw other coworkers on their back.

[gandolf989]

Originally posted by farrokhp
I agree that I might express that a little too strong when I said " all new features are dangerous to use in business" , That was not the main intention. What I am trying to say here is from manager point of view , assuming that you are a top notch DBA and you figure out these new stuff in a glimpse of a time, but if you left company and they can't hire another genius Jmodic!!!, then they are in trouble. So 9 times of 10 they decide to stay with core features. Also what I am saying is new features takes a while ( usually long time ) to become real practical solid features, I don't want to be victims of their bugs in the first place, doesn't mean they are useless but for the time being they are more marketing triggers to create more sales then eventually they will become either obsolete or real features.

If you write truly modular code then you can adapt to new Oracle features more quickly. Also if you write an abstraction layer between your code and Oracle supplied package you can 1) overcome any deficiencies in Oracle code and 2) isolate your code from any changes that Oracle might make to its code. I.E. If you need to write a work around to fix an Oracle bug in one version you can remove that work around when they fixe it in a later version. You can even put in a switch to decide if you need to work around a bug depending on the version of Oracle that the code is running in.

The bottom line to writing maintainable code is to write modular code where you don't repeat the same logic in multiple places.

Oh I guess I should add that IMHO in front of everything that I said. Including that part where I mention that you should not be so sensitive.

[slimdave]

Happily enough I can agree with all points of view here.

I couldn't run my data warehouse without bitmap indexes, partitiong, materialized views etc., but I do get nervous about new features.

For example we upgraded to 9iR2, and implemented bitmap index joins. Whoops! Hit a DBMS_STATS bug, and either had to abandon DBMS_STATS or the new index type.

I'd like to implement range-hash partitioning, but it's incompatable with MV Partiton Change Tracking. Even if it weren't, I' proabbly not use it because it's when you start combining usage of these new features that Oracle is likely to drop the ball.

[KenEwald]

I'd like to add my 2 cents to the discussion about OPS aka "RAC". Went to an Oracle Ed class about 2 yrs ago and realized that OPS is to be avoided at all costs (Oh, there's that too .. that stuff is expensive).

We've found hardware solutions that get us around OPS so far. Cheaper than licenses.

Some things like Oracle Internet Directory and Messaging are fairly new and work well.

[efshafi]

Yes I do agree with farrokhp views about oracle new features. Every time they release new patches after they found bug. Take for example data guard. In oracle 8i the data guard was not properly working the fail over and switch over is big hell. They claim in 9.2.0.1 it is very easy. The data guard latest patch no 2670975 & 2671349 dated 10.10.02 and last revision date 02.03.03 has to implemented then run these HIgh availabitlity utility. they also the dataguard manager wizard is very user friendly. But in relity it is not so. I am yet 2 implemen the same

[jovery]

I'd be interested in anybody could name a software house where thier latest features work perfectly first time. I've worked in this industry for many years and never found one, Oracle are not unique here.

Market demands mean they have to continually enhance their products to stay ahead of the competetion, unfortunately quality tends to take second place.