Deliverables - Performance Testing

Performance testing is a key part of delivering a Minimum Viable Product (MVP) as we build support for Fedora 6 as a back-end option. As part of the Working Group’s scope, we want to include performance benchmarking for key resource-intensive activities performed by Hyrax users. To reflect the current state of Hyrax adoption, the following combinations will be tested with representative performance tasks.

1. Hyrax, ActiveFedora, and Fedora 4 (for baseline comparisons)

2. Hyrax, Valkyrie, and Fedora 6

3. Hyrax, Valkyrie and Postgres

Representative Performance-Intensive Tasks

The following activities have caused known performance issues in Hyrax with ActiveFedora and Fedora 4.x. These issues are compounded when executed in a repository with a significant amount of content already ingested, but given the challenges of generating a representative content sample in a test environment, and trying to approximate a baseline operating environment, these test cases are suggested as minimum thresholds only. Fedora 6 test cases will also assume a basic Hyrax installation using a default Hyrax work type/model on a Nurax environment.

Additional References:

• Emory book ingests (works with over 100 Filesets)

• Emory large collection ingest rate (over 5,000 works)

• System Stack Error when deleting FileSets [is this still applicable?]

• Applying visibility change to a work with many FileSets in the UI fails/times out

• Avalon and Fedora 4-6 Performance Testing

Performance Tests (draft)

• Worksheet (please add more test cases, examples and/or local benchmarks)