BSD certification.

[Paul Robinson]

Mooneer Salem wrote:

>I meant to reply to this message, but I haven't had the time until now.
>:) Anyways, while reviewing for the LPI certification exams (there's a
>lot of obscure stuff that isn't used often in server administration, it
>seems) I thought of "the ultimate (insert OS here)" certification exam.
>
Wow. Haven't seen this thread for a while. It kind of died out as I got 
busy with stuff at work-work, but I do have some notes I'm collating. 
It's turning into a monster of a project here, but I'm having to step 
back a little here because I want to talk to some people who know about 
accreditation professionally. I work at a University in a project that 
is involved with defining new CPD and accreditation mechanisms, so I 
want to bring in some of that experience.

<snip>

>After taking the multiple choice section of the exam, the test takers
>would go to another room with rows of computers on the desks. Before the
>actual exam takes place, the exam creators would take a hard drive,
>install FreeBSD on it, and then basically manufacture a problem that
>would require a sysadmin to fix it (such as a trashed boot sector or
>Apache). Preferably the problem would normally take about an hour to two
>hours to solve. Each test taker would have access to a FreeBSD CD and
>the Internet (for Google). Network and routing information would be
>provided on a whiteboard in the front of the room in case the computers
>are not set up properly for networking.
>
OK, this is problematic. It causes problems because what you're 
proposing has implications:

1. Cost. What you're suggesting is expensive to setup, co-ordinate and 
run. I'm fed up of hearing about people not wanting to do CPD because 
it's a "waste of money". We can address that by increasing quality 
within our own domain, AND reducing cost as much as possible.
2. Consistency. How do we check the guys in CA are getting tests as hard 
as the guys in Afghanistan? Employ an army of people who can check the 
quality and standards of the testing centres?
3. Complexity. Is the ability to walk into a room and fix a trashed boot 
sector useful for somebody who is going to running an ISP operation 
where a trashed machine just gets re-installed? How do we manage CPD 
goals, skills, etc. into a framework that is useful yet adaptable.

Your thoughts are good and I know at least one person who has spoken to 
me off-list would like to get something like that together - maybe you 
should get in touch with each other - but this doesn't feel like it's 
going in the same direction as I am. I promise, I am working on this 
despite my quietness over the last month, and when the docs are ready 
for public consumption, I'll post the URL up here. I'm really keen to 
get some feedback.

Currently, the key salient points are:

- All testing done on-line as multiple choice. Sure, the guy can IRC to 
a guy who has done it, but his 100 questions will be picked randomly out 
of more than 2,000, and he only has an hour to complete it. You either 
know it, or you're going to cut and paste - the latter will take too 
long, and we'll take points off you for not finishing. That means costs 
are reduced (no testinf centre required), and we can maintain quality 
whilst retaining a sense of security.

- Tests produce points that build into larger groupings that produce 
qualifications. This is hard to explain, but the tests you do to get a 
qualification won't be the ones the guy next to you does. It means if 
you're into mail and dial-up, and the guy next to you is into web and 
dns, you can both get CPD qualifications, but you don't both have to 
know each others stuff. This gets mad - I know exactly how to do this, 
but space prevents me explaining here.

- This is not about FreeBSD. Sure, there'll be FreeBSD tests, but I want 
to produce something wider ranging that shows you have skills that 
employers, peers, whoever, will respect you for. FreeBSD will be in the 
first set of tests we produce, but I'm not going to ignore 
Open/Net/DragonFly/Darwin/Linux/QNX/whatever

- These tests are going to be hard. REALLY hard. You're going to have to 
study no matter how long you've been a Unix wizard. Sure, Greg Lehey 
might get the FreeBSD  kernel-hacking tests down on the first try, but 
he might be the only one who does. The point is, people should respect 
this stuff. I don't want to work on a Unix equivalent to the MCSE 
qualifications. My spare time is too valuable. I'm sure your training 
budget is too.

- Lastly, if you do a test on FreeBSD, it won't cost you much, but a 
percentage of that fee goes to FreeBSD. We might spot areas of docs that 
need filling in so will throw money at the docs project specifically (we 
want all study material to be freely available), but in general, if you 
pay for a set of tests on the BSDs, the BSDs will get money. We 
obviously need to take out an operational cost but I'll keep that as low 
as I can for now. This is something I haven't even started to talk to 
people about, but it makes sense. How many other ways can you think of 
convincing your employers to donate cash to FreeBSD? :-)

>BTW: Paul, maybe we can chat via private email more about your ideas for
>certification, if you'd like. :)
>
There will be a mailing list setup for this in the next couple of weeks, 
because lots of people want to get stuck in. If I drive this (I will, 
even if nobody else wants to get involved and even if nobody looks like 
they would sign up :-) ), it'll be run as a limited company in the UK, 
but I want as many people as possible (wellm, as many as is reasonable) 
to get involved in at least the planning and design, if not the 
implementation.

Let me get back to you all in a couple of weeks, and in the meantime, 
nohup & it to the back of my brain and let that work on it.

-- 
Paul Robinson