Introduction to SKA Regional Centres. Séverin Gaudet CADC

Transcription

1 Introduction to SKA Regional Centres Séverin Gaudet CADC

2 Outline The need for SKA Regional Centres The SKA Regional Centre Model The SKA Regional Centre Coordination Group Thoughts on a Canadian SRC 2

3 The need for SKA Regional Centres The main factors that drive the need for SKA Regional Centres (SRCs) 1. The science data products that emerge from the SKA observatory are not in the final state required for science analysis 2. The data volumes are so large that direct delivery to end users is unfeasible 3. The community of scientists working on SKA science data products will be geographically distributed 3

4 The need for SKA Regional Centres The main factors that drive the need for SKA Regional Centres (SRCs) 1. The science data products that emerge from the SKA observatory are not in the final state required for science analysis! Generation of advanced data products not in scope of project! Combination & further analysis of data products outside of observatory boundaries 2. The data volumes are so large that direct delivery to end users is unfeasible 3. The community of scientists working on SKA science data products will be geographically distributed 4

5 The need for SKA Regional Centres The main factors that drive the need for SKA Regional Centres (SRCs) 1. The science data products that emerge from the SKA observatory are not in the final state required for science analysis 2. The data volumes are so large that direct delivery to end users is unfeasible! Does not account for possible future "discovery" archive! Final data volume for each project will exceed that delivered by the observatory! Downloading data to local machines/cluster expensive and unfeasible in long term! "Take the processing to the data" 3. The community of scientists working on SKA science data products will be geographically distributed 5

6 The need for SKA Regional Centres The main factors that drive the need for SKA Regional Centres (SRCs) 1. The science data products that emerge from the SKA observatory are not in the final state required for science analysis 2. The data volumes are so large that direct delivery to end users is unfeasible 3. The community of scientists working on SKA science data products will be geographically distributed! KSPs with 1000s of hrs of observing time will dominate the science programme! Large international teams drawn from across the membership! Need new methods, algorithms and techniques! Driven by the community so they need a platform on which to do this 6

7 The SKA Regional Centre Model An international collaborative alliance of interoperating centres Transparent and location agnostic interface to SRCs for users SKA Regional Centre Alliance SRC 1 No SKA user should care where their data products are located All SKA users should be able to access their data products, irrespective of whether their country or region hosts a regional centre SKA Observatory SRC 2 SRC 3 SRC n SRC Science Gateway Users 7

8 The SKA Regional Centre Model An international collaborative alliance of interoperating centres Governance: How the SRCs are made to function as an alliance and to work serving the needs of the SKAO and the user community; Science Archive: Provision of an archive for the storage and curation of, and access to, SKA science data products; Accessibility and Software Tools: How users will interact with the SRCs; Data Processing: Generation and visualisation of science data products; User Storage: Different capabilities for users, including scratch spaces and project collaboration spaces Network connectivity: Transfer of science data products from the Observatory into SRCs and between SRCs 8

9 US$.1M/ Year US$.1-.2M/ Year US$1-3M/ US$1-2M/ Year US$.2-.5M/ Year US$.2-.5M/ US$.2-.5M/ Year Year US$.2-.5M/ Year US$1-3M/ US$1-2M/ Year US$1M/ Year US$1-3M/ US$2-3M/ Year US$.1M/ Year Fibre Cable Systems SKA relevant Major NREN Paths SKA relevant v First Guess at Distribution and Annual Costs 10 year contract per 100 Gbit circuit Guesstimate of Regional Centres based on SKA member locations

10 SKA Regional Centre Coordination Group Define minimum set of requirements for SRCs Define an accreditation process and draft MoUs between SKAO and SRCs Define process for ingestion of ingestion and curation for user-generated data products in the science archive Define individual and whole network, including links with SKA telescopes Members: Antonio Chrysostomou Chair (SKAO) Séverin Gaudet (NRC, Canada) Jeremy Main (SKA-SA) Peter Quinn (ICRAR, AUS) Yogesh Wadadekar (NCRA, India) Michael Wise (ASTRON, NL) Shenghua Yu (BAO, China) Externals: Ian Bird (CERN) Andy Connolly (LSST, UWash) Lourdes Verdes-Montenegro (IAA, Spain) SKAO: Rosie Bolton (SRC Project Scientist) Miles Deegan (HPC Domain Specialist) Nick Rees Software & Computing) 10

11 A Canadian SKA Regional Centre Archiving User support Functions/Activities User workflow infrastructure Project collaboration infrastructure Processing and storage Interoperability infrastructure Algorithm development Pipeline development Infrastructure software development Education and outreach Alliance governance A Collaboration/Partnership The building blocks already exist: Universities NRC Canadian Advanced Network for Astronomical Research Compute Canada CANARIE CFI Innovation and Cyberinfrastructure funding SRCCG representation SRC networking subgroup representation (CANARIE) 11

12 A Canadian SKA Regional Centre Functions/Activities A Collaboration/Partnership Unlocking the Radio Sky with Next-Generation Survey Astronomy Archiving The building blocks already exist: User! support Successful $10M CFI Innovation Fund proposal Universities (PI: Brian Gaensler) User! workflow Development infrastructure of this infrastructure will also NRC allow us to train the next generation of Canadian physicists, software developers and data Project scientists, collaboration and will infrastructure establish the capacity Canadian needed to Advanced host the Network Canadian for SKA Data Centre Astronomical Research Processing and storage Compute Canada Interoperability infrastructure CANARIE Algorithm development CFI Innovation and Pipeline development Cyberinfrastructure funding Infrastructure software development Education and outreach Alliance governance SRCCG representation SRC networking subgroup representation (CANARIE) 12

13 A Canadian SKA Regional Centre Challenges Understanding how users will use a regional centre Matching infrastructure to support user workflows Estimating processing and storage requirements Some numbers based on Canada s nominal SKA share of ~6% Storage: 36 petabytes at start; 60 petabytes/year growth Processing: 30 petaflops at start; 60 petaflops by 2030 Networking: 100 gigabit for SKA data alone Activities Continued participation in SRCCG and networking groups Developing collaborations with other SRC initiatives: AENEAS, Eridanus Building experience with existing projects: VLASS, CHIME, MWA, ASKAP, MeerKAT 13

14 14 14