hours sitting in a small biege room using nothing but a mouse I picked
up an 88% score and certification.
Now I have a headache. It
was tougher then I expected, I was hoping to have an educated guess at
the 25% of questions I wouldn't know. That's the great thing about
multiple choice. However this was evil multiple choice. Which two
answers is correct? Which three answers are correct? You've got to
guess right more then once on these questions.
Signing Up
I
went to the Prometric web site, picked IBM off the list, then found the
000-415 exam on the long list of IBM exams. Here is a tip, they are
sorted by exam number! There was a AUS$276 but IBM might reimburse me
under "You Pass We Pay" partner plan. I believe it is US$100 in the US
and prices in other parts of the world range between the two.
I
had four testing centres in Melbourne to choose from, each with
different days and different sessions, one test site (Monash Uni where
I graduated many years ago) would not load. I chose the only one with
Thursday afternoon exams.
Test Preparation
I booked an
afternoon test so I could spend the morning studying. That was really
dumb, I should have booked a midnight test to fit in the studying I
needed. I read up on the following doco:
- Parallel Job Advanced Developers Guide
* environment variables (APT_) they recommend you change.
* Buildops and custom stages.
- Parallel Job Developers Guide
* Read up on the partitioning types I never use (modulus).
* Took a quick look through the enterprise DB stages I never use.
* Tried to find some doco on orchdbadmin as I had never heard of it.
* Configuration files and node pools.
(That's the great thing about certification, it's filled in a few gaps).
- Server Job Developers Guide
* Refresher on the command line interface dsjob, dssearch and dsadmin.
- Designer Guide
* Ah ha, found the orchdbadmin in here.
* Read up on any sequence job stages I don't normally use.
* Make sure you know your stage variable uses
* Environment and project specific variables.
- Installation Guide
* Had a quick read of kernel settings as it's in the objectives. The test whammed me with a doosy of a kernel question.
* Read up on USS deployment as it's in the objectives though there was no sign of it in the test.
Taking the test
Lucky
I wore the brown underpants, three quarters of the way through and I
didn't think I would pass. Every second question seemed to be about
configuration files! There are a lot of developers out there who never
play with config files as it gets setup for them at installation time.
There also seemed to be an unholy obsession with sorting. Luckily I'd
spent a lot of my study time looking at the different ways a job can
sort and it certainly helped in the test.
There are 75
questions, most of them are multiple choice and at least half require
the selection of 2 or 3 correct answers from 4 or 5 provided. This
makes it harder to guess your way through. There were several good
questions with pictures, a couple with pictures of Director logs, a
couple with pictures of a job where you had to fill in the blanks. I
hope they can expand on those type of questions in future tests.
The ratio of questions matches the test objectives on the IBM website.
After
one pass through I felt like I was 50% right with forty minutes left. I
had marked about 15 questions I didn't think I had right. On the second
pass through I counted the questions I was 100% confident about and
came up with 19 questions that I felt I had an outside chance of
getting. I spent more time on those questions.
The final score
was 88%. I got 6 out of 11 database questions wrong, which accounted
for about 8% of my errors. So I did quite well in all other categories
with only one or two questions wrong.
What did I get wrong?
Didn't
spend enough time learning each of the enterprise database stages and
load options. I have a very good knowledge of DB2 Enterprise and a
passing knowledge of Teradata Enterprise. They just kept firing Oracle
Enterprise questions at me!!!
Was also too shallow on my knowledge of installation and configuration files.
There were three questions I didn't think I could get right as they didn't make sense.
Dodgy Questions
I
felt there were three questions in my set that did not make sense.
There was one on sequence jobs where I had to pick 3 of the 5 options
and any way I looked at it only two of them were right. And I know
sequence jobs quite well.
There were a couple questions that had
technical jargon that obscured the question and didn't relate well to
real life use of the product. I think some plain English could be
introduced into some questions. I feel sorry for the English as a
second language student.
It's this language barrier that makes
me think the visual design questions should be increased and the wordy
theory questions reduced.
The Wrap
Overall the test
was harder then I expected. I felt they ignored a lot of everyday use
stages and focussed a lot of optimisation (config files, sorting etc)
whereas I would like to see more job design questions.
Free Question
Okay, here is one of the exam questions, couldn't let you go without a sneak peak.
How many certified DataStage developers does it take to change a lightbulb?
a) I only know one and he's too lazy to change a lightbulb.
b)
Each developer can change one lightbulb per parallel node, with
combined operators turned on they can also change the smoke alarm
battery.
c) This joke started off lame and is now just downright embarrassing.
d) Two of the above answers are right but the other two are wrong.
Disclaimer: The opinions expressed herein are my own personal opinions and do not represent my employer's view in any way.