Revisiting the Aviation Safety vs. Networking discussion

Joe Provo nanog-post at
Fri Dec 25 11:09:54 CST 2009

On Thu, Dec 24, 2009 at 01:09:26PM -0500, Randy Bush wrote:
> > I _do_ create action plans and _do_ quarterback each step and _do_
> > slap down any attempt to deviate.
> imagine a network engineering culture where the concept of 'attempt to
> deviate' just does not occur.

Whimsical deviations don't belong in the maint execution, they belong 
in the brainstorming and design.  Gather more points of view during 
the peer review of the specification of work.  In my experience, good 
engineering makes for bad drama (and conversely if it is a "dramatic 
save" then you have a bad engineer and likely a cowboy).  Have a plan 
that executes in stages, tests at checkpoints where partial completion
is possible, and a fallback for each step.  A great way to train up 
junior people, document as you go, expose flaws and lines of future 
investigation, and if things go south you escalte to those who can 
judge *reasonable* new directions.

To me, that kind of change management for non-automatable work is a 
descendent of resonable group work.  If you have project-oriented 
autonomous teams that stick to the guideposts of "your standards" and
"minimal disruptions/maximal uptime" then good work will emerge.  As 
for automation, that enables your expensive hmans to do more smart 
things so should always be incorporated in processes and be something
people move toward, IMO.



             RSUC / GweepNet / Spunk / FnB / Usenix / SAGE

More information about the NANOG mailing list