First off: great to see you again! Welcome back! I would really appreciate to see you more often here.
Quote:
Originally Posted by
zazzybob
Whilst the few people in this thread may not use SAMBA, iSCSI or Kerberos, there are a lot of people that do, and they are still relevant skills to possess. If you want to be certified as a Linux generalist, you should know how to do these things.
This is actually not what i am complaining about in the modern certifications (i don't know the ones your organisation offers, so what i mean here is the likes of MCSE, CCE, CATE, etc. and yours might be different). What i am complaining about is the way these tests are designed: suppose you want to know if someone is proficient with, say, file system design. What the common tests do is to find out if the candidate knows every one of the 37
ls-commandline options.
Now, i ask you: regardless of knowing or not knowing these 37 commandline options: in real life the problem might be a broken hard disk, which needs to be rescued: companies like Ontrack can copy the data to another (working) disk, but are you able to reconstruct the content of the bad block so that the disk is readable again? Actually a colleague of mine and me did this once for an AIX disk, using
dd and
awk to reconstruct the FS metadata. It was not only factual knowledge: it was factual knowledge along with the imagination about how to apply that and the experience which told us how to approach such a problem in the first place.
Another example: take "knowledge of LDAP". It is all well and fine to know about the protocol and the tools and procedures, etc.. But this will only help you to create a badly designed LDAP domain if you haven't learned the ("intrinsic") knowledge of what sets apart a good from a bad design. Such knowledge ideally comes from experience: yours, when you do it and the experience of a "mentor" (doesn't matter how you call him) who guides you around possible pitfalls and passes to you what he has learned from his failures. All these latter mentioned things will not be tested by some multiple choice tests like "name the 3 methods of ...." a), b), c), d), e).
A "test" which would indeed test the worthiness of an expert (and not just how good he is qualified as man page) would include giving real-world-(like-)problems to people and judge the solutions they come up with, just like it is done at the academical level: get a theme, write a thesis, then defend it before peers. If someone wants to become a physician he has to do some multiple choice tests too - but ultimately he is given a corpse and has to show that he has what it takes to operate on people. And the task is not "name every tissue you see" but "do a [put your favourite operation here]".
I would be glad to see certifications be worth something: if a person sports an "MD" after his name i know i can trust him to treat me if i'm ill (well, granted, there are better and worse ones). I'd like to see all sorts of certifications giving me the same level of trust about whatever the area of expertise is that is certified.
But again, this is just a pipedream of an ageing hippie, unfit for modern business.....
bakunin