[Info-vax] Beyond Open Source

[johnwallace4 at yahoo.co.uk]

On Sunday, 10 May 2015 17:34:04 UTC+1, David Froble  wrote:
> Stephen Hoffman wrote:
> > On 2015-05-10 04:32:03 +0000, IanD said:
> 
> >> I'd love to see an event logger like windows has on OpenVMS :-)
> 
> Well, I personally find it severely lacking in necessary details.
> 
> > Microsoft very much understands the enterprise.
> 
> Yeah, they understand where the big bucks can be had, and they've 
> brainwashed enough people to make it easy to harvest those bucks ..
> 
> 
> What I don't understand is this aversion to independently implemented 
> procedures and such.  One size doesn't fit all.  If someone designs and 
> developes something that precisely fits their needs, why is this so much 
> worse than generic stuff?

Windows Event Logger lacking in details? [1]

Look, it's pointy clicky shiny, what more could you want, other than
graphs?

Get yourself some mentoring from a proper presentation layer person,
then come back. Heathen.


[1] Many years ago, working at DEC, I was dragged in to a customer
meeting at zero notice due to the storage rep being off sick.
Storage wasn't really my thing, but VMS and clusters were of mutual
interest, so we broadened the agenda.

Customer in question had decided to have fix-it services done by a
cheaper vendor than DEC. People were doing that at the time.

Customer mentioned an unexplained crash in their cluster (which
did 24x7x365 critical control of a shop floor) that their fixit
provider couldn't explain. I asked if they still had dump files,
error logs, operator logs and so on. They said no, they routinely
deleted such logs (not archived, deleted).

Despite that, I asked if they'd let us (initially, me) have a look
at the cluster. Reluctantly they said "yes, have a look, you won't
find anything".

Their procedures were sufficiently unrigorous (and VMS sufficiently
verbose AND rigorous) that some log files *were* still around. In
this case there were some unwelcome-looking mount verification
messages (amongst others).

Knowing I was out of my depth at that stage, I got the CSC folks
involved and from the residual information available they
identified a specific patch that should have been applied but
hadn't. Job done. Not just "it's Tuesday again, try these patches".

I would be astounded if anything like that happened with a so-called
"enterprise class" Window box and support service. Maybe it could
happen with Solaris, they used to seem to take reliability and
serviceability more seriously than Redmond, from what I've heard.

Some people would say that's not evidence, that's an anecdote.

I'd probably agree, except in the years since 1993 when I first
started using NT (and I haven't stopped since), I can't recollect
anyone EVER seeing any truly useful info in the NT event logger
(or in NT crash dumps, for that matter). I can recollect seeing
huge volumes of largely useless but nicely presented information
in the NT event log.

Some people might also say that the market has changed since then.
Well obviously it has. E.g. disposable hardware is allegedly
trendy, and where disposable hardware is a bit extreme, HYPErvisors
(almost inevitably on x86-64) are fashionable.

How well does this shiny work when it comes to diagnosing
unexplained misbehaviour? If you're a presentation layer operation
whose main purpose is serving ads and gathering user data
(Facebook, Google, etc), it doesn't exactly matter much if you get
the occasional unexplained failure or data corruption. If you're a
transactional operation or if there are (e.g.) lives at stake,
supportability might matter a bit more.

Or maybe I just don't get out enough.

"One size doesn't fit all."

;)