[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [dist-obj] Am I Insane?

   What type of bug would you say is more severe?
   (Note: I randomly ordered the two as to avoid biasing you.)

   1. Crash, Java Exception, hang
   A. Data Corruption, Data Loss

   The context is assigning of priority for bugs,
   both for pre-release QA testing, and for escallations.

   I have my opinion, others in... the company under discussion...
   have a contrary opinion.
My $.02 cents.

Well, w/ the information provided, it's hard to say. If you are
shipping product, and if the product fails to work w/ no customer
workaround.. either of these could be classifiable as priority 1 for
us.. which could affect our SLA's with our customers.. i.e. the clock
starts ticking when the application stops working.

If a customer has workarounds, or if these situations occur rarely,
and affect your product only marginally, well the priority can be
adjusted lower. (Note that a crash may obviously be much more "in your
face" and embarassing during demos or bake-offs, but a data
corruption/data loss situation could be much harder (read $$) to
recover from esp. if it's irreversible. Which is worth more to you --
preventing embarassment for your marketing VP during a trade-show, or
preventing your support staff from poring over backup tapes?)

Personally speaking, I've found Java exceptions and crashes easier to
debug, and if QA/testing happen correctly, should rarely occur in
production. Hang's are different matter (given Java stability) --
we've suffered thru those, and in distributed systems they can be hard
to repro/track down. In my experience, data corruption/data loss
always ends up sucking up more resources.. we attempt to mitigate that
risk through using Oracle (haven't lost a single record yet in years
of production systems), and through clustered solutions to improve
availability.  IMO, if you have the possibility for data corruption,
that's a far more dangerous/insidious situation that should be stamped
out in development.. costs for recovering from such situations once
you are in production can be astronomical sometimes. 

So, to summarize, I guess I'd vote for A. over 1. during development,
but the point I'd like to make is that you should define your
priorities in terms of what's important to your customers, not in
terms of functional descriptions of problem categories.

 

==========================================================================
To manage your subscription, mailto:dist-obj-help@distributedcoalition.org 
Archives, FAQ, etc.     http://www.distributedcoalition.org/mailing_lists/