35846a9b7c
In a lot of placeses we are using word "benchmark" which can mean workload, subtask, or test case which is very confusing. This patch partially address wrong usage of "benchamrk" word Change-Id: Id3b2b7ae841a5243684c12cc51c96f005dbe7544
1.1 KiB
1.1 KiB
Add support of persistence task environment
Use Case
There are situations when same environment is used across different tasks. For example you would like to improve operation of listing objects. For example:
- Create hundreds of objects
- Collect baseline of list performance
- Fix something in system
- Repeat the performance test
- Repeat fixing and testing until things are fixed.
Current implementation of Rally will force you to recreate task context which is time consuming operation.
Problem Description
Fortunately Rally has already a mechanism for creating task environment via contexts. Unfortunately it's atomic operation: - Create task context - Perform subtask scenario-runner pairs - Destroy task context
This should be split to 3 separated steps.
Possible solution
- Add new CLI operations to work with task environment: (show, create, delete, list)
- Allow task to start against existing task context (instead of deployment)