Cloud Dataverse
- 1. Cloud Dataverse Mercè Crosas (IQSS, Harvard University) with Orran Krieger, Piyanai SaowaraBtada, Ata Turk, Anuj Thakur, Gustavo Durand, Leonid Andreev MassachuseIs Open Cloud (MOC)Workshop, December 6 -‐7, 2016, Boston University
- 2. Dataverse IncenUvizes Data Sharing • A widely used open-‐source plaWorm for building data repositories • Gives essenUal incenUves to data authors: – get aIribuUon and credit through data citaUon – retain control over data published in the repository • Fosters a community to: – build new standards and best pracUces – increase research in data sharing
- 3. Dataverse repositories are installed in 21 sites around the world Harvard Dataverse repository: • 63,000 datasets; 12 new datasets published per day • 2 Million data downloads; 1,500 downloads per day • 15,000 registered users • 5000 data authors from 500 insUtuUons
- 4. Data depositor Data users Metadata Data files Data + metadata Access object in Swi` + Compute with Sahara/Hadoop download Swi` Object Store Dataverse Now with Cloud Dataverse Repository Publish dataset Data Replica3on
- 5. Cloud Dataverse will be available to any Dataverse repository Each Dataverse repository can choose to enable the Cloud Dataverse opUon
- 6. Dataverse + MOC are being expanded to support Cloud Dataverse 2016 Summer Pilot: ü Dataverse supports an external object store ü Data are replicated from a Dataverse repo to MOC Working on: q From Dataverse, user can access object in Swi`/S3 + compute with Sahara and Hadoop q Single authenUcaUon to data access and compute
- 7. Cloud Access + Compute This dataset has been enabled in Cloud Dataverse