Test Automation for Data Warehouses
-
Upload
patrick-van-renterghem -
Category
Data & Analytics
-
view
820 -
download
0
Transcript of Test Automation for Data Warehouses
Challenges
• Lead times of BI / Analytical projects are ever decreasing• Tools
• Automation (Wherescape...)
• Virtualization (Denodo, CIS...)
• Appliances (Netezza, DATAllegro...)
• Methodologies
• Data Vault
• Agile DWH
BI went Agile, but Testing didn’t
Challenges
• Data• Larger data volumes
• IoT
• Unstructured data / poor data quality• Social networks
• No relevant test data sets available
• The number of possible test cases are near infinite
Data Volume & Quality
Challenges
Review
requirements
Review
requirements
Define testing
Strategy
Define testing
StrategyPrepare Test DataPrepare Test DataEntry / Exit criteria
Design Test Cases &
Scripts
Design Test Cases &
Scripts
Configure the test
environment
Configure the test
environment
Prepare & Design
Agree on Entry / Exit
test criteria
Agree on Entry / Exit
test criteria
Integration
testing
Integration
testing
Performance testing Performance testing
Acceptance testing Acceptance testing
- Basic testing
- DI jobs accessible
- Reports accessible
- Cleansed
- Complete
- Correct
- Integrated
- Valid reports
- Relevant data
- Available data
- Consistent data
- Accessible
- DI & BI integration
- Full test cycles
- Check NFRs
- Scalable
- Check SLAs
- Peak user
- Peak loads
- Functional
- User
- Production
- SME
- End user
Construct
Defect metrics
review
Defect metrics
review
Performance
statistics
Performance
statisticsLessons learntLessons learnt
Process & Quality Improvement
Accept
Data Completeness
testing
Data Completeness
testing
BI & Analytical
Testing
BI & Analytical
Testing Smoke test
Unit test
Smoke test
Unit test
BI Testing Lifecycle
Challenges
** Regulatory compliance might require to use the v-model, e.g. validated environments.
Functional
analysis
Requirements
functional
Non-functional
Technical
Design
ConstructionUnit
Smoke Test
Data
Completeness
BI &
Analytical
Integration
testing
Performance
User
ProductionAcceptanceValidation
Verification
BI Testing V-Model
Complete DWH Testing
• Complete testing is a must
• It requires:• Testing methodology
• Right project culture/mindset and organization
• Tools
Our View
Complete DWH Testing
Database
integrity
checking.
Risk based
testing
Effective defect
management
and
collaboration
End to End
Performance
testing
Adherence to
compliance and
regulatory
standards
and…AUTOMATE
Critical Success Factors
Test AutomationLeverage social development principles to deepen functionalities
Testers
- functional testing
- regression testing
- result analysis
Developers / DBAs
- unit testing
- result analysis
Data Analysts
- review, analyze
data
- verify mapping
failures
Operations teams
- monitoring
- result analysisCo
lla
bo
rati
on
/ W
ork
flo
wTe
st M
an
ag
em
en
t
Rational Quality Manager JIRA Team Foundation ServerQuality Center
Test Automation
Basic functionality
• Auto detection of anomalies (or at least prior to being detected by a user)
• Targeted regression testing for planned changes
• Data error identification errors by comparing (huge) data result sets
• Data error detection via a rules engine
• In between data layer reconciliation and auditing
• Test data generation
Additional functionality
• No programming or coding
• Heterogeneous connectivity
• Collaboration and workflow capabilities
• Visually attractive development and monitoring environment
• Intuitive reports & dashboards
Functionality
Test Automation
• Shorten regression cycles
• Save report developers time
• Test the same data set in less time
• Test more
• Faster deployment of defect resolution cycle
• Faster deployment of enhancements
• Less cumbersome upgrades/migrations
• Enabler for• Implementing continuous testing• operationalization of testing
Benefits
Test Automation
Vendors - tools
• ICEDQ
• RTTS -QuerySurge
Unconnected ETL
Source Data Layer Target Data Layer
ETL - ELT
1 3
2
Load test data
Execute Job externally
Extract result
4 Compare & report
• Validation of the business rules implemented in ETL processes are assessed by comparing the results against a ground truth.
• The ETL processes are executed separated from the test automation tool
• Less adequate for multi-step ETL processes.
Approach
Test Automation
Vendors - tools
• Zuzena
Connected ETL
• The business rules present in the ETL tool are analyzed and the test results are assessed against the anticipated results.
• The test automation tool executes the ETL processes.
Approach
Source Data Layer Target Data Layer
2 4Load test data Extract result
5 Compare & report
1
Analyze logic
3
Run job
Test Automation
Vendors - tools
• Report Valid8tor (BO)
• 360Bind (BO)
• Integrity Manager (MSTR)
• Report Validator - BSP Software (Cognos)
Report Integrity validator
• Parallel testing of 2 live systems
• Comparing against a (historical) ground truth
• Comparing against known good baselines
Approach
Reports
Reporting data Layer
Load test data Scrape data
4 Compare & report
1 3
Run report
2
QuerySurge
• Integrates with HP QC / IBM RQM / MSFT TFS
• Provides collaboration features
pulls data from data sources
pulls data from target data store
compares data quickly
generates reports, audit trails
reports
SQL
Design Tests
Scheduling
Reporting
Run
Dashboard
Wizards
Data Health
Dashboard
• a SQL execution and data comparison tool working against heterogeneous datasources.
• No programming needed
Automade introduction
• Automade• Provides an agile answer to the ever-increasing information appetite
• By automating the mind-numbing aspects of constructing, maintaining and testing of data warehouses.
• Automade is a spinoff of MindThegap and has established partnerships with
Test automation
Thank you for listening,
Any questions?
Feel free to send questions & feedback to [email protected]