Designing User Interfaces Spring 1999

1
SE 767-NT Software Performance Engineering Robert
Oshana Lecture 30 For more
information, please contact NTU Tape
Orders NTU Media Services [phone]
oshana_at_airmail.net
tapeorders_at_ntu.edu
2
Performance testing

• Confirm software performance meeting objectives
• Are models representative of software behavior?
• Test individual scenarios
• Test various load conditions in the proper
  environments

3
Baseline models

• Models evolve
• Use CM to baseline models for tracking and
  reporting the performance evolution of the
  software, and for further investigation of
  earlier results

4
Post deployment performance management

• Evolutionary changes
• Design instrumentation to provide data for
  quickly and easily updating models after software
  development
• Capacity management
• This is an established discipline and part of the
  computer system management organization

5
SPE artifacts

• Four broad categories
• Plans
• Performance objectives
• Performance models and results
• Performance validation, verification, and
  measurement reports

6
SPE artifacts

• Performance Management Plan
• Performance VV plan
• SPE CMP
• Performance Drivers
• Performance Scenarios
• Performance Objectives
• Execution environment specifications

7
SPE artifacts

• Performance models
• Model results
• Performance instrumentation
• Performance VV reports
• Performance test plans
• Performance test results

Adopt the artifacts to the organization
8
Integrating SPE into the SW process

• Value of a well defined software process is
  recognized
• Integrating SPE into the software development
  process is not that difficult
• Define the right milestones and deliverables

9
Waterfall model

• Classic Waterfall does not really work
• A modified Waterfall can work in certain projects
• SPE should begin during analysis
• Models during product concept
• Evolved during design/implement
• Performance testing

10
Spiral model

• Real projects seldom proceed in sequential
  fashion
• Iteration required between phases
• SPE fits naturally iterate on
• Establishing objectives
• Identify and resolve risks
• Performance measurement

11
Unified process

• Supports OO and UML
• Iterative and incremental
• Mini projects
• Inception
• Elaboration
• Construction
• Transition

12
Unified process

• Key attributes for SPE
• Architecture centric
• Use case driven
• Risk driven
• Integration of SPE and UP is straightforward

13
SE 767-NT Software Performance Engineering Robert
Oshana End of lecture For
more information, please contact NTU Tape
Orders NTU Media Services [phone]
oshana_at_airmail.net
tapeorders_at_ntu.edu