Re: [SystemSafety] Software reliability (or whatever you would prefer to call it)

From: Yiannis I Papadopoulos < >
Date: Tue, 10 Mar 2015 11:50:15 +0000


"I object to the use of the term "reliability" being applied to anything other than failures (using the term loosely) resulting from physical degradation over time. I believe it is important to maintain a clear distinction between undesired behavior designed into a system, and undesired behavior that arises because something ceases to function according to its design. (Here "designed / design" is used broadly. It includes all intellectual activities from requirements to implementation.)"

Dear Michael,

But from this objection it follows that you seem to expect every measurement of reliability to refer *only* to failures caused by physical degradation of the hardware that fails.

For the objection to hold, you effectively must assume that every other possible cause has somehow been excluded from the measurement, i.e. either that the system has been designed developed programmed manufactured and installed *correctly* or that we are certain that the failure has *not* been caused by a "design fault" (in the broad sense). The latter can include not only software bugs but also things like an imperfect electrical connection in a motherboard caused by temporary fluctuation in temperature during production. Must one also exclude bit flips caused by particle radiation from the causes of failure that count in reliability measurements? I think this is an impossible assumption to make in practice.

We might as well not bother about reliability estimation from measurements at all if we adopt this assumption.

kind regards

Yiannis Papadopoulos -

http://www2.hull.ac.uk/science/computer_science/our_staff/staff_profiles/yiannis_papadopoulos.aspx



To view the terms under which this email is distributed, please go to
http://www2.hull.ac.uk/legal/disclaimer.aspx



The System Safety Mailing List
systemsafety_at_xxxxxx Received on Tue Mar 10 2015 - 12:50:20 CET

This archive was generated by hypermail 2.3.0 : Tue Apr 23 2019 - 02:17:07 CEST