Fri Afternoon Notes

Dec 9, 2011 Fri Afternoon Notes

Scalability/Production Deployment

  • Stanford SDR approx. 200 k objects preservation; DOR access copy 200k
    • Object design refactoring from Inline XML
    • How to deal w/ lots of parallel write? e.g. from scaling up workers
    • expect millions when legal issues are resolved to allow ingest of Google content
  • Columbia >100k but not sure
    • Rails caching for scaling access
  • Hull 5k going to 20k
    • mostly reads, barring initial ingest/migration
  • UVa "hundreds of thousands of objects"
    • HSM for archive, Netapp storage for Fedora access objects
    • interested in leveraging cloud storage and perhaps running Fedora in the cloud
  • Eddie gave anise overview of the possible redundant and scalability options for a current project.
    • general need for ability to scale up read nodes
    • general need for best practice around backups, consistency audit
    •