BenchmarkQA Performance Testing Quality Forum March 2012
-
Upload
benchmarkqa -
Category
Technology
-
view
1.201 -
download
0
description
Transcript of BenchmarkQA Performance Testing Quality Forum March 2012
Performance Testing Fundamentals for Managers
Presented by:
Jeff RothVP of Consulting Services
BenchmarkQA
Date:
March 22, 2012
Agenda
• Performance Testing 101
• Components of Performance Testing
• Performance Testing Lifecycle
© 2012 BenchmarkQA Slide 2
What is Performance Testing?
© 2012 BenchmarkQA
LoadVolume
Stress
ScalabilitySystem Benchmark
Durability
System TuningDesktop Benchmark
FailoverSystem Optimization
Baseline
Infrastructure Validation
Batch Jobs
Failure Point
Slide 3
Business Goals• Will my system meet my goals?
– User expectations (response time, accuracy)– Load capacity (user volume, cost of configuration)– Data integrity (at normal loads, peak loads, and failure)– Stand up over time (durability)– Failover capability (continuity through failure) – System growth (scalability)
• When will my system degrade and when will it fail?
• What do I need (infrastructure) for effective performance?
© 2012 BenchmarkQA Slide 4
Please excuse me while I shout….
PERFORMANCE TESTING IS DEFINED BY WHAT YOU WANT TO ACCOMPLISH
NOT WHAT TOOLS YOU OWN
© 2012 BenchmarkQA Slide 5
Before we answer this question, let’s start with a little anatomy of IT systems…
What Can Performance Testing Do?
© 2012 BenchmarkQA Slide 6
• Input – Keyboard, Touch Screen, Voice, Camera, Hand Scanner, etc.
• Output – Screen, Printer, Sound, etc.
• Processing – CPU, GPU
• Storage – RAM, Disk, SDD, etc.
• Connectivity – LAN, WAN, WiFi, 3G/4G
User Interfaces
© 2012 BenchmarkQA Slide 7
Performance Testing Devices
Business Goals
• Configuration Optimization– Best hardware for the job– OS/Software configuration
• Baseline Device Performance– Embedded devices – software dependent upon hardware– With and without connectivity
© 2012 BenchmarkQA Slide 8
Performance Testing Systems
• Include devices or not
• Real vs. simulated integration points
• Isolate risk areas
• Iterative process
• Load testing tools…
© 2012 BenchmarkQA Slide 9
COMPONENTS OF A PERFORMANCE TEST
© 2012 BenchmarkQA
What Makes up a Performance Test
© 2012 BenchmarkQA
Web Farm Mainframe
Databases
Target of TestLoad Generator
Controller
Slide 11
What Makes up a Performance Test
Virtual User Script
• Activity/Business process
© 2012 BenchmarkQA
Search ItemHome Page Select Item
Slide 12
What Makes up a Performance Test
Technical Details of a Virtual User Script
• Protocol - HTTP, AJAX, SOAP, SAP, CITRIX, etc.
• Uniqueness– Session variables– User IDs– Test Data
• Files• Correlate within the script• Made up
• Think time
© 2012 BenchmarkQA Slide 13
Performance Test Scenario
• One or more test scripts
• Quantity of VUsers per script
• Velocity of activity per user
• Designate ramp up & test schedule based on goal
© 2012 BenchmarkQA Slide 14
Ramp Up And Failure
© 2012 BenchmarkQA
Errors begin to occur @ 400VUsers
Number of VUsers AxisPercent of
Errors Axis
Slide 15
PERFORMANCE TEST LIFECYCLE
© 2012 BenchmarkQA
© 2012 BenchmarkQA
PMBOK - PROJECT MANAGEMENT PHASES
SELECTION INITIATION PLANNING EXECUTION & CONTROL CLOSURE
Project Definition Requirements Analysis & Design
Technical Analysis & Design
Build/Integrate & Test
User Acceptance Test
Deployment & Wrap-Up
Design
Build
Execute
Analyze
Initial Planning InitialPlanning
Slide 17
Softw
are
Dev
elop
men
t Life
cycl
e (S
DLC
)Pe
rfor
man
ce T
estin
g Li
fecy
cle
Performance Testing Lifecycle
Design
Build
Execute
Analyze
Initial Planning
The Performance Test
lifecycle requires some initial
upfront planning work, and
becomes a cyclical effort to
achieve your performance
goals.
Typical engagements are 7 to 9 weeks including initial planning.
© 2012 BenchmarkQA Slide 18
Participants
ØProject StakeholdersØProject ManagerØSystem ArchitectsØKey SMEsØPerformance Architect
Duration
1 to 4 Weeks
4 Test Effort Scope4 Software Components4 Target System Architecture4 High Volume Business Processes and Resource
Intensive System Activities4 Workload Profiles4 Performance Test Goals
4 Resources4 Target Test System Hardware4 Project Personnel and Test Effort Roles4 Test Data Management Plan4 Test Tools (load injection hardware and software)
4 Test Schedule4 Project Schedule4 Project Dependencies4 Performance Test Work Breakdown Structure
© 2012 BenchmarkQA Slide 19
Participants
ØBusiness SMEsØSystem SMEsØPerformance ArchitectØPerformance
Engineers
Duration
1 to 4 Weeks
4 Load Test Tools4 Test Script IDE4 Load Generation Farm4 Protocol Licensing (Web, Winsock, Citrix, etc.)
4 Business Process4 Screen by Screen Walk Through4 Usage Frequency and Pacing4 Test Data & Test User IDs
4 System Intensive Processes4 Batch Jobs4 Reporting Systems4 Interfaces/Feeds
4 Test Monitoring Strategy4 Servers/Databases/Network4 Integrate with Load Test Tool (Security
Permissions)4 Data Center/ SME System Monitoring
© 2012 BenchmarkQA Slide 20
Participants
ØBusiness SMEsØSystem SMEsØPerformance
Engineers
Duration
2 to 12 Weeks
4 Test Scripts4 Record Business Processes4 Correlate Data4 Parameterized Data4 Pacing4 End User Response Time Measurements4 Typically 4 to 8 Scripts
4 Test Scenarios
4 User Ramp Up Scenario
4 Frequency of Business Process Activities
4 User Managed Activities
4 Results Storage
4 Verify Test Readiness
4 User Concurrency
4 Multi-Script Concurrency
4 Load Generators© 2012 BenchmarkQA Slide 21
Participants
ØProject StakeholdersØProject ManagerØSystem ArchitectØKey SMEsØPerformance
Engineers
Duration
1 to 8 Weeks
4 Schedule Execution4 No Users in Target System4 Anti-Virus Scans, Automated Backups4 Off Hours Scheduling
4 Monitor Execution4 Load Test Progress4 Load Test System4 Real Time Response Time Metrics4 Target System Utilization Under Load
4 General Execution Approach4 Small Load First To Confirm System Is Stable4 Reasonable User Load Ramp Up4 More Than One Test Execution To Confirm
Goals
© 2012 BenchmarkQA Slide 22
© 2012 BenchmarkQA
Typical User Response Graph
Slide 23
Participants
ØProject StakeholdersØProject ManagerØSystem ArchitectØBusiness SMEsØSystem SMEsØPerformance Architect
& Engineers
Duration
1 to 4 Weeks
4 Test Goal Assessment4 Response Time Performance4 System Utilization
4 Bottleneck Identification4 Correlate Response Metrics with System
Utilization Metrics4 Isolate Causal System/User Activity
4 Remediation Strategy4 Configuration Changes4 System Changes4 Hardware Changes4 Test Script Changes4 Load Scenario Change4 Monitoring Changes
Repeat Cycle Until Goals Achieved© 2012 BenchmarkQA Slide 24
Things to Remember• Set clear objectives
– To ensure all team members agree on the questions performance testing will answer
• non-functional requirements• solution requirements specification
– To understand the expectations of the business owners– To know when testing is complete
• Strive for realism– Subject the system under test to conditions that match production as
closely as possible– If possible use the actual production hardware for the performance test
© 2012 BenchmarkQA Slide 25
Things to Remember• Understand what happens
– Every component of a system must be monitored and data collected– Subject matter experts analyze collected data
• Provide meaningful recommendations– Must improve the end-user experience– Must be based on a cost-benefit analysis
© 2012 BenchmarkQA Slide 26
Thank You!For additional information on Consulting Services with
BenchmarkQA, please contact:
Jeff Roth952.392.2384
© 2012 BenchmarkQA