1 of 9

University of Chicago

PI Chats

Rob Gardner

August 25, 2022

IRIS-HEP Focus Areas

1

2 of 9

Who is actively working on the project from the group and what are they doing

  • Rob Gardner
    • SSL coordination, facility R&D, support for workshops & training events
  • Suchandra Thapa (Software Professional)
    • ServiceX development & integration
  • Lincoln Bryant and Fengping Hu (DevOps Engineers)
    • SSL services, K8s infrastructure (Flux, IdM, etc), development and production clusters, Coffea-Casa testing, integration with ATLAS, production AF, monitoring
  • Leveraged efforts from UC team: physical facility infrastructure, ServiceX DevOps and Rucio integration, portal training, user support

2

3 of 9

What are contributions of those individuals in your group to the various area(s), i.e. AS, IA, DOMA, SSL, SSC, OSG-LHC, MGMT

  • Providing development and production SSL resources
    • SSL-development (river-dev), SSL-production (river), ATLAS Tier3-production (af.uchicago) clusters
    • Implemented with Analysis Facility R&D principles and concepts resulting from blueprint meetings and workshops
  • We are actively working with AS team on providing feedback on Coffea-Casa development
  • In DOMA, we are a leading contributor to ServiceX development, ATLAS integration, deployments, functional and scalability tesitng
  • With AS and DOMA teams, building and operating infrastructure to support full pipeline testing, AGC tools workshops, and ultimately at-scale AGC demonstrators
  • For SSC, deployed CoDaS-HEP notebook and ML resources
  • For MGT, working to align IRIS-HEP goals with US ATLAS Operations

3

4 of 9

What specific collaborations has the group made with other groups within IRIS-HEP?

  • DOMA ServiceX, Skyhook
    • Illinois, Washington, Santa Cruz
  • Analysis Services
    • NYU, Illinois
  • OSG-LHC
    • Wisconsin
  • SSC
    • Princeton
  • SSL
    • UNL (consulted on analysis server specs)

4

5 of 9

What specific collaborations has the group made with other groups or individuals external to IRIS-HEP?

  • DOMA ServiceX
    • Texas, Fermilab
  • Analysis Services
    • Brookhaven, Fermilab
  • ATLAS Distributed Computing Analytics
    • Frontier-Squid
    • PerfSONAR analytics
  • PanDA-Harvester team
  • FAB (FABRIC Across Borders)

5

6 of 9

Which contributions has the group made to the "Intellectual Hub" aspect of IRIS-HEP? (Including community building activities, training, outreach and broader impact.)

  • Presentations at ATLAS Software and Computing workshops
  • Analysis Facility
    • working with ATLAS on analysis facility definition
  • US ATLAS
    • 'Shared Tier3' development with other US ATLAS sites (SLAC, Brookhaven)

6

COLLABORATIONS

​

  • OSG-LHC, SLATE�(containerization, deployments)�
  • DOMA (deployments, functional testing, scale testing)�
  • CoDaS-HEP instructors, US ATLAS Ops�
  • US ATLAS Physics & Software Support�
  • ATLAS ADC (Analytics deployments on SSL), PanDA (Harvester containerization)

​

7 of 9

Please list any papers/presentations/software-repositories connected to the above.

  1. Presentations
    1. 2 Aug 2022 - NSF Shared Analysis Facility at UChicago, Fengping Hu, Annual US ATLAS S&C Technical Meeting, UMass Amherst
    2. 23 May 2022 - New Advances in Analysis Facilities, Analysis Ecosystems Workshop II
    3. 25 Feb 2022 - Update on UC Shared Tier3, Lincoln Bryant, US ATLAS Meeting
    4. 16 Mar 2022 - "Kubernetes at UChicago: PATh, IRIS-HEP Scalable Systems Lab, U.S. ATLAS, and SLATE", Lincoln Bryant, OSG All-Hands Meeting 2022
    5. 4 Oct 2021 - ServiceX, Related Tools, Demo, Ilija Vukotic, ATLAS group meeting
  2. Software repositories
  3. Services and websites
    • https://af.uchicago.edu/
      • ServiceX - xAOD
      • ServiceX - uproot
      • Coffea-Casa

​

​

7

8 of 9

Are metrics aligned with the project? Do metrics accurately reflect the progress, success or impact of the project?

Metrics are�well aligned�

​

8

Metric

Target

Current Value

M.6.1: Number of deployed and operated platform services

Design phase targets taken from initial DOMA and AS estimates. A suitable metric for the execution phase will be desgned.

12

M.6.2: Number of SSL user groups

Determined by number of developers & other devops requiring resources

>9

M.6.3: Number of engaged institute areas per quarter

5 Institute areas were targeted for engagement

4 (DOMA, AS, Training, OSG-LHC)

M.6.4: Number of integrated SSL cluster resources

Determined by scale demands requirements of development teams, both within the Institute and outside

2 (1 dedicated, one shared)

9 of 9

How is your project managed? How is progress measured? How are risks identified and mitigated?

  1. Management
    1. Daily contact through Slack
    2. Brainstorming design sessions
    3. Bi-Weekly meetings (DOMA, ServiceX)
    4. GitHub issue tracking, SX PRs and releases managed by Ben Galewsky
    5. Weekly group meetings
    6. Many other meetings - regular and out of band we're asked to join
  2. Progress
    • SSL - c.f. Metrics (app deployments, IRIS-HEP groups engaged, etc.)
    • ServiceX - currently contributing to 1.
  3. Risks - examples
    • SSL - technology risk (Kubernetes)
      1. LHC community adoption, pushback by Facility incumbents
      2. Lack of contributions, expertise, effort
      3. Culture of isolation, island projects
    • ServiceX
      • scalability - testing & measurement
      • community adoption, physicist interface
      • integration with PaNDA/iDDS (as backend "executor") - has been delayed due to lack of expertise & effort

9