UC Berkeley Cloud Meetup 036: High Performance Computing (HPC) Cloud costs

This month the Cloud Meetup will host a presentation and discussion by Gary Jung (Gary Jung is the lead for the Scientific Computing Group at the Lawrence Berkeley Lab and is the manager for the High Performance Computing service at UC Berkeley), and the resident architect and engineer over Savio - Berkeley's high performance computing service.

Talk Details:
UC Berkeley's cloud strategy has been called "University first" - rather than emphasizing the tool over the purpose and outcomes. That said, we are always looking on how to improve and employ new technologies. There are major barriers to using cloud computing (such as Azure, GCP or AWS) for traditional high performance computing. Come to the meetup to learn about what HPC is, and the specialized kinds of challenges Gary and the team have to address. Gary will share data on the costs for HPC, and some side-by-side comparisons with getting equivalent resources from public cloud vs traditional colocated servers. We will also explore how things could evolve in the future-- especially on finding the right tool for the job and thinking about cloud more as a way of working rather than as a destination.

Everyone is welcome - this is a really inclusive group! Even when the talks seem very technical you don't have to be an expert, or already know about cloud technologies; we get together so people can learn from each other.

CONNECT VIA ZOOM: Our Zoom connection info changes each month and is posted here in meetup invite. [Note: for security reasons attendees must be logged into a Zoom account]

About the presenters:
Gary Jung is the lead for the Scientific Computing Group at the Lab and is the manager for the High Performance Computing service at UC Berkeley. He has been in the HPC field for over 20 years and started the Scientific Computing Support program at the Lab in 2003 — one of the first such programs at a large Tier 1 research institution. He also led the effort to launch Lawrencium in 2008, a Top500 list HPC Linux cluster at the time, in order to make institutional HPC available to all Lab researchers. Other projects include a strategic role in the development of the highly successful Berkeley Research Computing program at UC Berkeley in 2014.
Gary holds a bachelor’s degree in Computer Science from San Francisco State University. His hobbies include traveling, cooking and working on HPC (high performance cars).

AGENDA:
1:00pm - welcome, announcements and poll
1:15pm - Presentation & Q&A on HPC in the cloud