...
Agenda
- Topics on fedora performance/scaling
- 2016 work summary
- Review Performance and Scalability Test Plans
- Graphing of JMeter results, script
- I/O benchmarking of test environment, script
- Collection of basic test environment characteristics, script
- Hydra - Fedora performance (?)
- Islandora - Fedora performance (?)
- Next meeting time 02/13/2017 11am.
Minutes
Topics on Fedora Performance/Scaling
- Nick - none
- Colin - mocking large-scale tests, alt-methods of doing that
- desire to test multi-TB loads
- Kieran - interested in testing processes and procedures
- are both Tomcat and Jetty tested? - mostly Tomcat
- Kevin - interested in validating release candidates
- now more generally interested in loading issues
- multi-threading topics
- Danny - interested in scaling as it relates to HyBox
- "many members" issue
- multi-tenant support
- Colin has done this in Hydra... but has not been able to backup single hierarchies
- using different base-containers
- Yinlin - interested in large files
2016 work summary
- Are the defined tests adequate for the interests above?
- Challenge in have a test environment that has enough disk space
- Maybe use OpenSSH:rand to mock many files
- Is running this on AWS infrastructure a possibility?
- Who is going to pay for a 20-TB EBS volume?
- Performant disk i/o can be expensive in AWS
- AWS tests should include data around size/time/cost of previous tests
- We need to summarize the results of 2016 testing
- Colin available to help in the visualization of the previous test results
- Yinlin has created graphs of his tests
- Colin has modified scripts based on team feedback
- Yinlin - need to write report for each test
- Interest in maintaining results in a GitHub
- ACTION: create results repo in fcrepo4-labs
- ACTION: write a report from last year's results
- Current reports do not provide immediate clarity of the state of Fedora
- Volunteers:
- Colin to work on scripts - fcrepo4-labs
- Nick/Yinlin to work on drafting - will use GitHub
- Colin to generate R-Script outputs that can be starting point for summary doc
- Data result locations: post to S3
Actions
- Colin Gross to transfer fcrepo_perf_analysis to fcrepo4-labs
- Andrew Woods to post test outputs to S3
- Colin Gross to push outputs from R-scripts to GitHub
- Nick Ruest and Yinlin Chen to take above R-script output as starting point for performance test summary report