Re: Total oracle newbie

On Wed, 12 Feb 2003 15:17:11 +0000, Jeff wrote:

[snip]
> I have not had too bad an experience with the instructors I've had. Few
> complaints overall. Probably the worst was when I felt like an instructor
> was theory-only (teaching-only) without sufficient hands-on working
> experience--curriculum experts--which sometimes made answering my
> questions harder or less profitable. But, good or plodding, I think I
> usually got what I needed from the courses I took.

There's always that danger, of becoming 'classroom-bound'. In Oz, certainly, there is precious little scope for instructors to 'take time off' and go spend, say 3 months every 18 months, in (say) consultancy. So yes, it can happen. It helps to minimise the problem if the instructor can  listen as well as talk.

>
>>The people who write the course material, by the way, are all undoubtedly
>>extremely competent. But that doesn't mean they don't have to skate over
>>some issues which are capable of sparking a religious war if subject to
>
> As I've taken my share of courses (both ILT and CBT), I've seen the errors
> in the course materials, and these were not just semantically or arguably
> wrong, but patently, blatantly wrong.

Well, mistakes will always happen. But I can't think of too many complete and utter howlers. I do my fair share of 'well, that's not right', but it's arguable that on many occasions they are getting a 'gist' of the truth across and not wanting to get bogged down in sticky details.

I agree it's not 100% perfect, but considering the quantity of stuff that's produced, it's not outrageously innacurate.

>Thankfully, instructors usually
> pointed them out so that we could correct them; although, I have had to
> point out a few myself. Practice exercises that wouldn't work,

Well, that's not entirely a fair criticism. I'm sure that if every training centre was configured identically, and all configured the same as the writers' setups, many of those glitches would disappear.

>examples
> that were illogical, etc. These things should've been caught if someone
> took the time to review/test/edit the material before distribution.

Remember, too, that many of the examples etc are based on beta software, because the course material has to be available shortly after the release to market of the production code. I remember, for example, a classic slide in the 9i New Features course, explaining the new 'WITH' syntax: it just didn't work as written. But it did in the beta, so my puzzlement was explained. Remember, finally, that the writers are working when the software is extremely new: sometimes they can only go on what the developers tell them it ought to/will do when the thing hits the market!

I'm not making excuses, by the way. And I'm not saying things couldn't be better. But there *are* review processes, and course material, like software, still manages a few 'bugs'.

>
>>close debate (witness what goes on here on certain topics, such as the
>>right block size or the need to separate indexes from tables): courses
>>last a finite number of days, and the material has to fit accordingly!
>>Sometimes that means writing things which are 'generally' true, but not
>>always specifically accurate. If you get an instructor who's willing to
>
> I've seen a great deal of strong condemnation here of Oracle's Tuning
> curriculum. Are you now saying that there's not significant room for
> improvement there?

Absolutely not. But how should they go about it? There are two principle schools of thought on tuning. The Ratio Tuners do it one way, and the Wait-Staters do it another. The RTers have history on their side; the WSers have inifintely more thoughtful, accurate and effective theory on theirs. Oracle's running with the history crowd at the moment. It's safer that way. And it's not actually "wrong", just not terribly insightful.

There's a lot of material on database structures and coding -that's all highly relevant. Chapter 14 starts with the magisterial (and totally accurate) statement that most tuning isn't (or shouldn't be!!) done by DBAs but by designers and developers. Personally, I find it a bit of shame they leave that nugget of wisdom to chapter 14, and kick off instead with Library Cache Hit Ratios and the like. Which is why I teach the chapters in what I'm sure looks like a totally random fashion to some students (We start at Chapter 14 and finish with Chapter 6!!). But that's just me, and the fact that I'm a WSer.

>I think it's pretty bad that what Oracle teaches about
> performance tuning is considered largely a waste of effort by real-world
> DBA's (and some Oracle instructors).
>

I wouldn't say it was a waste of effort, but it's true (for me) that it doesn't get to the heart of the matter. But it would be a profound ideological shift to do it any other way, and there would be a whole bunch of other criticisms you could make if they tried doing it any other way. What we've got is a common denominator course -which is OK as far as it goes.

>
>>look into some of those subjects, it's a bonus, not a sign that the
>>material is necessarily bad. For the same reason, not all instructors get
>>to preview the material before release because it would never be released
>>if they did!
>>
>>And also, speaking from personal prior experience, not everyone who is
>>good at delivering training is good at writing (or advising on) training
>>material.
>
> True. But surely a company with Oracle's resources and manpower should be
> able to find enough of those who are to improve the course materials to a
> point where it isn't such an easy target for criticism. What of
> instructors that have actually offered Oracle Education help in this area
> and, seemingly, those offers were ignored or rejected?

I've no knowledge of that happening.  

> Oracle training, IMHO, is generally good, don't get me wrong. It just
> sometimes seems to me like OE isn't doing all that it should to make the
> training better... such that students need never question the value of
> their Oracle training later.

I put it this way: it's not really called 'OE' any more, but OU (Oracle University). And real Universities have research departments. And Oracle should/could do with one too. And then I think a lot of the valid stuff you've mentioned would not be an issue any longer. As it is, it's 3.30am in Sydney right now, and once these posts are out of the way, it's back to some testing... ie, instructors who want to understand what they teach in anything like a profound-ish way have to do their own research when they can. There's no provision for instructors not to be instructing, in short.

Regards
HJR Received on Wed Feb 12 2003 - 10:30:40 CST