Research Automation for Data-Driven Discovery
Download as PPTX, PDF1 like182 views
This document discusses research automation and data-driven discovery. It notes that data volumes are growing much faster than computational power, creating a productivity crisis in research. However, most labs have limited resources to handle these large data volumes. The document proposes applying lessons from industry to create cloud-based science services with standardized APIs that can automate and outsource common tasks like data transfer, sharing, publishing, and searching. This would help scientists focus on their core research instead of computational infrastructure. Examples of existing services from Globus and the Materials Data Facility are presented. The goal is to establish robust, scalable, and persistent cloud platforms to help address the challenges of data-driven scientific discovery.
Research Automation for Data-Driven Discovery
- 1. Research Automation for Data-Driven Discovery Ian Foster Argonne National Laboratory & The University of Chicago foster@anl.gov
- 2. A productivity crisis in research Data volumes are growing much faster than Moore’s law … (10,000x more over 6 years for genome data) Kahn, Science, 331 (6018): 728-729 But most labs have extremely limited resources Heidorn: NSF grants in 2007 < $350,000 80% of awards 50% of grant $$
- 3. "Well, in our country," said Alice … "you'd generally get to somewhere else — if you run very fast for a long time, as we've been doing.” "A slow sort of country!" said the Queen. "Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that!" The challenge of staying competitive
- 4. 4https://bit.ly/2l4gfgu How industry handles complexity
- 5. cloud4scieng.org Industry software builds on powerful platform services
- 6. Cloud platforms have transformed how software is developed and delivered 6 Can we do the same for science? • Identify cross-cutting capabilities required by many groups • Define simple REST APIs for accessing those capabilities • Operate high-quality, scalable, secure, performant cloud-hosted implementations • Ensure persistence and evolution over time In so doing, enable many scientists and tool developers to automate and outsource tasks that are not central to their core mission: thus reduce costs, increase quality, promote interoperability
- 7. What capabilities? 7 • Auth: Manage identities, authentication, and authorization • Transfer: Manage movement of files from A to B • Sharing: Manage who can access data at a location • Publish: Preserve, identify, describe, curate • Search: Index and search data • Identifiers: Assign identifiers to collections of files • Automate: Organize sets of activities • Learn: Discover, train, run machine learning models • …
- 8. globus.org Science services operated by UChicago for researchers worldwide
- 10. Automate and outsource with REST APIs and Python SDK
- 11. Automate and outsource with REST APIs and Python SDK 11 UK NIST NSF NSF NSF DOE NSF Canada
- 12. Automate and outsource: Publication and discovery Move to permanent location (or publish in place) Compute and record checksums Obtain and record metadata Assign persistent identifier Index for discovery 1212 Data Publication Indexing materialsdatafacility.org 2 petabytes 100 Gbps Globus APIs
- 13. Automate and outsource: Publication and discovery 1313 Programmatic access (Python, Jupyter) Web browse and search Data Publication Indexing materialsdatafacility.org 2 petabytes 100 Gbps Globus APIs
- 14. Example: NCAR’s Research Data Archive Globus used for • Single sign on via streamlined account provisioning • Data sharing • Data downloads
- 15. 15
- 17. Cloud platforms have transformed how software is developed and delivered 17 We can do the same for science • Identify cross-cutting capabilities required by many groups • Define simple REST APIs for accessing those capabilities • Operate high-quality, scalable, secure, performant cloud-hosted implementations • Ensure persistence and evolution over time In so doing, enable many scientists and tool developers to automate and outsource tasks that are not central to their core mission, to reduce costs, increase quality, promote interoperability
- 18. We have identified some needed capabilities 18 • Auth: Manage identities, authentication, authorization • Transfer: Manage movement of files from A to B • Sharing: Manage who can access data at a location • Publish: Preserve, identify, describe, curate • Search: Index and search data • Identifiers: Assign identifiers to collections of files • Automate: Organize sets of activities • Learn: Discover, train, run machine learning models • … Established 12,000 endpoints 100,000+ users New 100s of users Experimental 10s of users globus.org — Ian Foster — foster@anl.gov