Re: Certification - Mailing list pgsql-advocacy

On Fri, Nov 6, 2015, 2:58 AM David Fetter <david@fetter.org> wrote:

On Wed, Nov 04, 2015 at 03:09:38PM +0000, Sameer Kumar wrote:
> On Wed, Nov 4, 2015 at 5:44 AM Jaime Casanova <
> jaime.casanova@2ndquadrant.com> wrote:
>
> > On 29 October 2015 at 08:52, Bob Lunney <bobl@aweber.com> wrote:
> > > Ah,  the old "certification" fallacy!
> > >
> >
> > while i agree a certification doesn't say much... i also know that
> > people is continously asking me for certified people or
> > certifications...
> >
> >
> Yes, I get similar request whenever we take up a new project. Because they
> think certified professionals can deliver a project very well. I don't
> blame them. Here in ASEAN (specially in Indonesia, Malaysia and Thailand) I
> have seen botched-up implementation by self-claimed experts/freelancer for
> PostgreSQL.

If you think it's different elsewhere, I have bad news, or good news,
if knowing you're not alone in this situation makes you feel better.
People who engage these people without checking references and doing
their own due diligence are taking risks.

This would be true in the presence of a certification as well.

I'm not against making another certification program, but let's be
very realistic about what such a program could and could not do.

Complete or even plausible protection against incompetence and malice
aforethought is NOT among the things a certification could accomplish,
and you would be doing a disservice to the idea by selling it as that.

Even assuming a rigorous, task-based test with good fuzzing and
frequent updates, I see no proposals on the table for ensuring that
the person who takes the test is the same one who presents the
credential.  Any such proposal simply starts a technology war between
people wanting such credentials without being qualified and that
system.

Then there's the question of retention.  Performing well on an
examination one day (or week, or however long it takes) is not the
same as long-term mastery of the subject.  A person can, and many
students do, cram for a test, do well on it, and fail utterly to apply
the knowledge three weeks later, let alone three years.

> Earlier someone mentioned about the non-popularity of EDB certification in
> ASEAN.  I am not sure of history, but in recent past (24-30months time) we
> have seen a lot of people taking up EDB's PostgreSQL training and
> certification. This has gone hand-in-hand with higher adoption of
> PostgreSQL and EDB in ASEAN. And as more organisation take up PosgreSQL
> more people will take up certification exams. Even professional look up to
> certification as a value add in their resume and so certification authority
> should really check their merit before awarding it.
> Also I have seen cases of large BFSI and telecom segment customers where
> they want to make sure that they have certified staff before promoting a
> new technology to be used for critical system. This is again where I have
> seen learning path/training and certification plays an important role.

I'm curious about situations you've seen that involve free software,
especially software where there isn't an authoritative vendor.  It's
one thing to certify on, say, Red Hat.  It's a very different thing to
certify on Linux as a whole, a situation much more comparable to the
PostgreSQL one.

I am saying that any training needs to be backed by a learning path or curriculum. This is where training comes in. Candidates who would want to apply for certification should know what are they going to be tested for. The learning path prepared then for it. Saying you will be tested against "Performance Tuning" skills is open ended. That is where a proper training or a learning path in form of some guide book can be helpful. The learning path has to make sure that it does not end up being a formal education session / textbook. I feel this necessity would be there for any technology certification free or not. That's is what I pointed out as a "good" about EDB certification backed by a learning path.

Even if I as an Organization decide to opt for Postgres for some critical project I would want my DBA to kwarn and prove that he can take care of this DB. This is again where certification and what is that DBA certified against helps.

> I feel while certifications are really important both to back the adoption
> and also to promote more adoption, it has to be backed by a proper training
> or learning path. At the same time a lot of effort has to be spent in
> maintaining the level of scrutiny of skills.
>
>
> *Disclaimer:* My employer is a training partner and master partner to EDB
> and I am myself a certified trainer. My opinion is my own but is based on
> the experience of delivering training to about 150-200 candidates during
> more than 15 training sessions in last 2-2.5 years of time.

As Josh mentioned, training is a business, and if you do it right, you
can do a lot of good with it as a business.  I'm not seeing a
compelling argument here for having the PostgreSQL project itself
divert scarce resources toward supporting this business, though.
Getting awesome new software out the door each year is already
stretching those resources thin.

May be I chose different words to describe what would be needed but I too had the same reasoning that it will be difficult for Community to take it up (certification and training) given what it needs. It would be upto individuals to take it up in their own personal capacity or upto commercial entities to drive it.

Cheers,
David.
--
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778  AIM: dfetter666  Yahoo!: dfetter
Skype: davidfetter      XMPP: david.fetter@gmail.com

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate