January 20 2022

Agenda

• Introduction and welcome
• Draft Workgroup Charter
• Workgroup communication resources on GitHub
• Brainstorm on group priorities: what is the best use of this workgroup?
• Meeting frequency
• Topic ideas for the next workgroup meeting

Notes

Introduction

• Will not do individual member intros, but hope to get to know people as time goes on
• Acknowledge the CMAS Center's role in setting up the WG communication infrastructure

Draft Charter

• Steve walks the group through the charter
• WG not for time-sensitive technical support

Workgroup Communication Resources

• Discussion on technical issues from GitHub issues vs CMAS Discourse
• Need to see how Git Hub Issues end up being used in this WG
• Suggestion of using Issues to document code issues and/or documentation, technical questions should go to Discourse
• May be a way to transfer issues from GH to Discourse

Group Priorities

• What do you expect to get out of this workgroup? What's important for us to focus on?
• Dave Healy (NH DES): transferring files, moving away from using hard drives between modeling centers; how can different agencies better access the same set of files and work more collaboratively without the physical intervention of mailing drives; is there a way to work in real time; will be more of an observer, but hopes to bring the discussion back to colleagues in the region
• Byeong Kim (GA EPD): designing next version of modeling server, has been looking into virtual machines being used by cloud providers; sharing system designs, questions about egress fees and data sharing costs; developing and sharing AMIs; plugging into people experience using cloud services for computing ** Steve points on data sharing: EPA puts data into AWS S3 buckets, charge only comes when you download data from AWS, but there is no charge when moving data from S3 to AWS cloud computing instances
• Weining Zhao (TCEQ): modeling group is exploring cloud computing, high priority is the initial set up; would like to know from other users how they've set up the cluster in the cloud, and share scripts for submitting jobs, lower learning curve; data sharing: need to better understand costs of moving data around
• Margaret McCourtney (MN PCA): CAMx modeler with limited computing capacity; was able to run larger jobs (source apportionment) using AWS; once the server reaches end of life, should work on moving to the cloud; state has contract with Azure, and hopes to learn how to optimize the set up to get the runs done as quickly as possible with the least amount of cost
• Sarav Arunachalam (CMAS): should be poll to see who are existing members of the cloud, what models they're using; Zac to share poll results with the group to help direct efforts
• Gil Grodzinsky (GA EPA): MOVES modeling in the cloud, and SMOKE-MOVES to allow states to use this resource and not just EPA
• Kevin Briggs (CO CDPHE): barriers procuring HPC resources through their internal IT dept, run into problems even running AERMOD because of how IT is managed in their agency; mostly use consultants for this SIPs but have ambitions to do it in-house; move some existing datasets into the cloud, e.g., WRAP modeling data

Meeting Frequency

• Do we meet monthly or less frequent?

Next/Future Meeting Ideas

• Post in the forum about
• Compute times and costs
• Presentations on use cases/highlight cases
• Presentations on tips/tricks modeling in the cloud
• Should we hear from vendors?
• What systems/vendors are available? Cloud providers or consultants who have set up modeling in the cloud (package vendors)