This document summarizes Strider, a test automation framework for Hadoop applications built by InMobi. Strider allows for sanity, functional, and regression testing of Hadoop applications in an automated way using JSONs. It is built on top of Apache Falcon, Oozie, and Hadoop and provides features like bootstrapping test environments, generating and handling input data types, easy test creation, and extensibility. Strider helps drive 10x efficiency improvements over manual testing by automating tasks like test setup, execution, validation, and input/output handling.
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men 🔝malwa🔝 Escorts Ser...
Strider_InMobi_HadoopSummit_2015_Brussels
1. Driving 10X Efficiency In Testing Hadoop
Applications
Presenters : Swamynathan S & Karan Kumar
Company: InMobi Technology Services
2. Key Takeaways on Strider
• Test Automation Framework
• Built on Falcon APIs, Oozie, Hadoop
• Sanity, Functional, Regression Testing
• Bootstrapping
• Data type handling using JSONs
• Easy test creation
• Extensible
3. InMobi and Data
• InMobi - Ad Network
• Handles 1B users
• Crunches 5 TBs of data/day and its growing!
• Data pipeline orchestration - Apache Falcon
5. Feed 1Feed 1 Feed 2Feed 2
Process 1Process 1
Feed 3Feed 3
Feed 4Feed 4 Feed 5Feed 5
Process 2Process 2
Feed 6Feed 6
Process 3Process 3 Feed 7Feed 7
An Example of pipelines
6. Feed 1Feed 1 Feed 2Feed 2
Process 1Process 1
Feed 3Feed 3
Feed 4Feed 4 Feed 5Feed 5
Process 2Process 2
Feed 6Feed 6
Process 3Process 3 Feed 7Feed 7
Feed xFeed x Feed yFeed y
Process PProcess P
Feed zFeed z
Process Addition
7. Feed 1Feed 1 Feed 2Feed 2
Process 1Process 1
Feed 3Feed 3
Process 3Process 3 Feed 7Feed 7
Feed xFeed x Feed yFeed y
Process PProcess P
Feed zFeed z
Process Delete
8. Feed 1Feed 1 Feed 2Feed 2
Process 1Process 1
Feed 3Feed 3
Process 3Process 3 Feed 7Feed 7
Feed xFeed x
Process PProcess P
Feed mFeed m Feed zFeed z
Feed
Add/Delete
9. Feed 1Feed 1 Feed 2Feed 2
Process 1Process 1
Feed 3’Feed 3’
Process 3Process 3 Feed 7Feed 7
Feed xFeed x
Process PProcess P
Feed mFeed m Feed zFeed z
Feeds Update
10. Challenges solved resulting in 10X efficiency
Challenge Pre-Strider Way Strider Way
Validate Distribution Individual Script Automated and Seamless
Input Creation / Versioning
Non-scalable and
redundant
Standardised & Versionable
Validation Non scalable pig scripts Generic JSON Validation
Host setup Manual Automated
Regression Expensive Inexpensive
Random and Prod Data Manual Automated and Evolving
Parallelization Not straight forward Yes
End to End test No Yes
CI Not efficient Yes
12. Functional/Regression Test Creation
Suite
testng.xml
Suite
testng.xml
Test SetupTest Setup
Test ExecutorTest Executor
ValidatorValidator
H
D
F
S
F
A
L
C
O
N
F
A
L
C
O
N
O
O
Z
I
E
O
O
Z
I
E
Output
to
JSON
Output
to
JSON
Tester’s Role
JSON
to
Input
JSON
to
Input
Strider’s Role
InputInput
GoldSetGoldSet
InputInput
GoldSetGoldSet
Test
1
Test
2
17. Road Map
• Enhanced Data Generation
• Performance testing and metrics
• Automated way to generate golden set
18. Summary
• Test Automation Framework
• Built on Falcon APIs, Oozie, Hadoop
• Sanity,Functional and Regression Testing
• Bootstrapping
• Data type handling using JSONs
• Easy test creation
• Extensible