Loading…
Attending this event?
Learn more about Samvera Connect 20203 on the conference wiki page, including hotel reservations and information about fun things to do in Philly!

How to register for the conference: Sign Up or Log Into Sched to access the tickets and to build your schedule. Find the options to access your account at the top of the screen. Once logged in, click the purple “Reserve Tickets” button.

Call for proposals for lightning talks and posters is open until September 15th!
Back To Schedule
Wednesday, October 25 • 11:30am - 11:55am
Solving a System Migration Mystery

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Tweet Share
Problem: For the American Archive of Public Broadcasting migrating data from an old metadata management system (AMS1) to a Hyrax-like (AMS2) became too slow after more than 50K assets. To date we have tried replacing batch ingests with Bulkrax, improved the templates that build the system, conducted speed tests, and added additional AWS resources. The work has been a collaboration between SoftServ and GBH on Bulkrax and the data model, the Picard Maneuver, and an ID rewrite.

Results: Performance improved with Bulkrax, we reran failures only instead of the whole ingest, the Picard Maneuver not working. However a problem with IDs resurfaces and running a script to rebalance the tree. The unbalanced tree with the IDs is the smoking gun. Currently trying to ingest metadata into a post-gres backed app and hoping the speed improves are once the tree is balanced.

Will discuss lessons learned.

Speakers
avatar for Karen Cariani

Karen Cariani

Exectuive Director GBH Archive, WGBH Educational Foundation
The WGBH Medial Library and Archive and the American Archive of Public Broadcasting have thousands of hours of audio-visual materials. I am interested in systems to preserve this material, and make it as accessible as possible in a digital form/on-line, and using computational tools... Read More →

Wednesday October 25, 2023 11:30am - 11:55am EDT
Ballroom A
  Concurrent Presentation
Feedback form isn't open yet.

Attendees (5)