Applying Model-based Distributed Continuous Quality Assurance Processes to Enhance Persistent Software Attributes

Time and resource constraints often force developers of highlyconfigurable quality of service (QoS)-intensive software sys-
tems to guarantee their system’s persistent software attributes
(PSAs) (e.g., functional correctness, portability, efficiency, and
QoS) on very few platform configurations and to extrapolate
from these configurations to the entire configuration space,
which allows many sources of degradation to escape detec-
tion until systems are fielded. This article illustrates how
distributed continuous quality assurance (DCQA) processes
help improve the assessment of these PSAs across large QoS-
intensive software system configuration spaces. We also il-
lustrate how model-based DCQA processes enable developers
to run formally-designed screening experiments that isolate
the most significant configuration options, such as different
workload parameters, operating system, compiler flags, fea-
ture sets, and/or run-time optimization controls. Our empir-
ical results show that DCQA processes can be used monitor,
safeguard, and enforce PSAs at an acceptable level of cost and
effort.