Re: Assessment question on indexes

Comments embedded

<xhoster_at_gmail.com> wrote in message
news:20050210130700.814$b0_at_newsreader.com...
> "Richard Foote" <richard.foote_at_bigpond.nospam.com> wrote:
>> A typical example of an OCP joke of a question.
>
> I have to disagree with you here, Richard. I think most DBAs spend a lot
> of timing dealing with developers, managers, or other people that barely
> have a grasp on what a database even is. Therefore, the ability to
> extract
> meaning out of poorly worded questions, and the ability to make reasonable
> inferences in the face of ambiguity, are very valuable skills for a DBA
> to have.

Hi Xho,

So are you suggesting that OCP questions should typically be vague, imprecise and erroneous so that OCPs are skilled at dealing with vagueness and inaccuracies in real life scenarios ? Interesting concept !!

Can you imagine a doctors exam, with the following question:

Patient has a high temperature. Are they suffering from:

1. Viral infection
2. Spending too much time in a steam room
3. Tonsillitis
4. None of the above

And the correct answer given by the *medical profession* is A because B can't be right because most people don't stay too long in stream rooms and C can't be right because the patient in question doesn't have tonsils ...

The question would be regarded as a "joke" because it makes a number of naive "assumptions" and doesn't cater for other logical situations outside the vague scope of the question and given answers.

For the inexperienced and for those that depend greatly on so-called learning materials, the possible incorrect assumption might be that high temperatures are only caused by viral infections.

If there were sufficient such questions, then perhaps the exam itself might be drawn into some criticism, as well and the quality of those that gain "Professional" status by only passing such exams.

You hopefully get my point ...

>
>> Firstly, the wording of the question is not clear as to whether or not
>> the two columns are separate indexes, a concatenated index or either, nor
>> is it clear whether the index if concatenated must be ordered one way or
>> the other.
>>
>> *Assuming* they mean a concatenated index on (PRODUCT_ID, WAREHOUSE_ID),
>> then A could be true as Oracle may possibly use a full index scan, B
>> could be true as it could use an index skip scan path (as it's a 9i exam
>> question),
>
> It seems unlikely to me that product_ID would have low enough cardinality
> for a skip-scan to be useful.

But how can you possibly say that based on the information (or lack thereof) in the question. The company may only make a handful of products, you just don't know.

An index skip scan may be unlikely but then again ...

>
>> C could be true as it must update the index with the new
>> values (meaning that the index is "used" by Oracle in an modify index
>> entries sense)
>
> I think is reasonable to expect someone to interpret "use", in context,
> so as not to include maintenance due to DML.

I too would interpret the question in the way you have but then again, others may not so it could still be opened to some interpretation.

> Although the fact they only
> insert 3 values make me wonder if the index is any skinnier than the table
> itself, so maybe the FFIS for A) would not be very reasonable.
>

Exactly. The query may/may not use an index, it depends. As it could with B. As it might be interpreted in C. As it might not in any case and so give D the tick.

I'm not sure that OCP exams should be the environment in which to promote vagueness and inaccuracies.

Thankfully, most other "Professions" agree :)

Cheers

Richard Received on Fri Feb 11 2005 - 00:58:05 CST