> -----Original Message-----
From: Dave Harvey [mailto:xxxxx@syssoftsol.com]
Sent: Monday, April 29, 2002 9:07 PM
To: NT Developers Interest List
Subject: [ntdev] Re: Philosophical Rant [was Re: Writing
Drivers in Ja va]“Peter Viscarola” wrote in message
> news:xxxxx@ntdev…
> >
> > “Roddy, Mark” wrote in message
> > news:xxxxx@ntdev…
> …
> > > Until vendors care enough to make sure that competent programmers
> > > are producing reliable drivers, nothing is going to change.
> > >
> >
> > But don’t you see: Fast moving markets make this impractical. Even
> > when you know what you’re doing, you need a LOT of test
> time. Well,
> > at least I do, maybe your code is different.
> >
>
> (Mark writes his in C++…)
Sure, my code has bugs. Good processes can catch most of those bugs early -
like before they become a customer experience. ‘Fast moving markets’ are
indeed a problem, but so is performance. I pulled the 20% number out of the
air, choosing a value that was big enough that nobody would say that it was
unrealistic as a point where the trade off between reliabilty and
performance tipped to performance. The industry rule-of thumb in the fault
tolerant/highly available systems world has been and continues to be
something like 5%: customers will accept a 5% performance degradation for
5-9’s of reliability.
As far as I know there is no architectural solution to the problem of poor
software quality that does not incur an unacceptable performance
degradation. The solution to poor software quality is, surprisingly enough,
to improve the quality of the software.