Bake-Off True Confessions

Bake-Off True Confessions

Bake-Off True Confessions

Chris Wellens chrisw at iwl.com
Fri Aug 9 15:56:35 EDT 1996


Many companies have asked me why InterWorking Labs was not at the
bakeoff.  We were planning to attend.  Then two terrible things
happened.  (1) we moved to a new building and everything went wrong --
phones, Internet connection, etc.  in fact, we are still only half way
moved in.  (2) our new revision 4.0 was buggier than we anticipated and
one week before and two weeks after the bake-off, we were buried with
support calls.   This had never happened before, but then again, it was
the first major re-architecture of our product.  When the choice came
down to send an engineer to the bake-off or apply all resources to the
support emergencies, well it was clear we had to support our current
customers.  It was extremely unfortunate timing, but business is
sometimes like that.  We will be at the next bake-off.


I have some observations that are intended to be productive and
make the next experience more useful.


(1)  No one was in charge with total responsibility for the
success of the event.  If there had been, I think a lot more
bases would have been covered, as described below.


(2)  There were no stated goals.  Interoperability for printers
must be defined by the PWG;  the issues are completely different
for printers than for other networked devices (e.g. routers).
If you are going to declare that a device is interoperable, you
have to define the terms.  Otherwise how do you know when you
get there?  How do you measure success or failure? 


(3)  There were no measures of effectiveness.  After the fact,
everyone agreed that finding bugs is always useful.  Well true,
but if you go through the trouble and expense of getting this
august group together for a few days to do this, you want to
thoroughly and systematically find all the bugs that you can.
This requires a plan.


(4)  There was no plan with a schedule for the testing (e.g.
test conformance first, interoperability with NMS' second,
performance under heavy traffic loads third, application
interoperability fourth, pathological packets from NMS' fifth,
etc.


(5)  The tests that were conducted were trivial.  Although they
yielded some useful results, no conclusions about conformance,
compliance, and interoperability can be drawn.    


(6)  Results were made public. Whether or not the participants agreed
to this, it was not necessary and has the effect of discouraging future
participation from those who were on the fence and worried about public
disclosure.  The way this is usually handled is to publish something
like "this paragraph of the spec was unclear; three implementations
interpreted it this way, and seven did it that way", then a decision is
made on what route is right, and the draft clarified to reflect that,
and the feedback given to the draft editor and IETF.  


(7)  As far as a public demonstration of Printer MIB at Interop
goes, it seems to me this group has to get back to their
marketing counterparts with a plan.  Since Interop-Atlanta is
just a month away, you probably want to have some agreed upon
demo that you will all run, you probably want to get some
distinctive "tent cards" you can display in your booths that say
something like "Our printers are now SNMP-manageable", or
something like that, and finally, you probably want to define 1
or 2 key messages about why this is useful to a network manager.
You might also want to issue a press release with these 1 or 2
key messages and listing all the companies who are shipping
products or planning to ship products in the next 90 days, as
supports of "SNMP-manageable printers".


Finally, yes I promise to be at the next one, and put my "money
where my mouth is"  :-)  


Chris Wellens
InterWorking Labs, Inc.
WWW:  www.iwl.com



More information about the Pwg mailing list