Hardware Reference
In-Depth Information
4.1
Methodology
The first part of the evaluation organises these QATs into different development
phases. The process tailoring information (e.g. phase, artifacts) captured by
QATF is used to incorporate QATs into relevant development process phases.
Safety and performance activities are used to describe the common purpose of
using these QATs. The second part of evaluation organises QATs into different
risk categories in Section 3.3 according to their aims in risk management. Safety
and performance techniques are again used as examples for this evaluation.
4.2
Results of Evaluation
Table 1 shows some examples of the safety QATs and their fit into development
process phases. For example, FMEA is suitable for subsystem hazard analysis
(SSHA) and system hazard analysis (SHA) during the architecture and design
phases. This QAT is not appropriate for hazard identification or preliminary
hazard analysis (PHA) in earlier phases because it is intended to help to analyse
potential failure causes and their effects. Corrective actions will be recommended
to the potential hazards based on an assessment of their criticality. Detailed sys-
tem information and descriptions are needed in order to perform this QAT.
Table 2.
Organising Performance QATs into Different Software Development Phases
Development
Phase
Performance Activities
Performance Techniques
Requirements
Define and Analyse Performance
Requirements
Execution graphs (EG), Use Case
Maps (UCM), Layered Queueing
Network (LQN)
Performance Prediction
Performance Assessment for Soft-
ware Architecture (PASA), LQN,
Performance Evaluation Process
Algebra (PEPA), Stochastic Petri
Nets (SPN)
Architecture
Performance-oriented design Prin-
ciples and Patterns
Principles (e.g. Centering Princi-
ples, Shared Resource Principle),
Patterns (e.g. Fast Path, Batch-
ing)
Identify Performance Antipatterns Antipatterns (e.g. Excessive Dy-
namic Allocation)
Performance Prediction LQN, Markov Chain
Performance Principles Principles (e.g. Locality Principle,
Parallel Processing Principles)
Identify Performance Antipatterns Antipatterns (e.g. Circuitous Trea-
sure Hunt)
Design
Coding
Performance Solutions
Performance Patterns (e.g. Fast
Path, Batching)
Performance Testing and Measure-
ment
Load Test, Instrumentation (e.g.
ARM, Paradyn), Benchmark
Testing
Performance Enhancement
Performance Tuning
