The use cases and expectations that fall under the category of Fedora "performance and scalability" are diverse and evolving. It is therefore important to establish and maintain a set of reproducible benchmarks that will be run over various configurations of Fedora releases. 

Team

• Andrew Woods (DuraSpace)
• Yinlin Chen (Virginia Tech)
• Nick Ruest (York University)
• Colin Gross (University of Michigan)
• Danny Bernstein (DuraSpace)
• Trey Pendragon (Princeton University)
• Longshou Situ (University of California, San Diego)
• Kevin Ford (Art Institute of Chicago)
• Joshua Westgard (University of Maryland)

Project Plans

1. Performance and Scalability Test Plans (Update to Fcrepo 4.7.x and later version)

Benchmark Categories

Resource Scale

1. Large datastreams (i.e. binaries, non-RDFSources)
2. Multi-TB datasets
3. Large number of objects (i.e. containers, RDFSources)
4. Many members

Performance Characteristics

1. Ingest rates
2. RDFSource property update rates

Prior Results

1. Performance Testing
2. Performance Summary
3. Assessment Plan - Performance
4. Test 1 Results Summary
5. Test 2 Results Summary
6. Test 3 Results Summary
7. Test 4 Results Summary
8. Test 5 Results Summary
9. Many Members Performance Testing

Project Tools 

1. Fcrepo performance analysis: https://github.com/fcrepo4-labs/fcrepo_perf_analysis
2. Fedora 4 Ansible: https://github.com/VTUL/fcrepo4-ansible
3. Many member testing scripts: https://github.com/dbernstein/fcrepo-performance-test-scripts

Publication

• Towards Use And Reuse Driven Big Data Management. JCDL 2015 
• Performance of a Cloud‐Based Digital Library. ICADL 2015
• Evaluating Cost of Cloud Execution in a Data Repository. JCDL 2016

Presentation

• Open Repositories: Presentations

Workshop

• Code4lib 2017 workshop: Performance and Scale Testing of Fedora
• Fedora Camp NYC - 28-30 November 2016
• All Fedora workshops: Events

Other Tools

1. JMeter
2. Grinder
3. https://blazemeter.com/ - commercial service for running our JMeter tests