Data-intensive computational techniques have become indispensable in virtually every domain of science. The sheer quantity of data being generated by various instruments and devices presents a significant challenge for even the most advanced computing centers. Traditional off-line “batch” approaches to data analysis often times cannot keep pace with real-time streaming data. At the same time, an explosion of new software tools has given computational scientists an unprecedented number of quality choices for analyzing their data.
At the Texas Advanced Computing Center, we are helping scientists analyze data in “real-time” with the tools of their choosing using an event-driven, container-based approach to the actor model of concurrent computation. In this talk we will describe the actor model and the benefits of containers for isolated, reproducible, polyglot environments. We will also introduce abaco, a Python system we have developed in support of this effort and will describe some of the science use cases driving its adoption
After completing a PhD in Mathematics from the University of Michigan, Joe moved to the University of Texas where he has been building distributed systems, web services and analytic tools for a variety of scientific applications. He is currently a research scientist at the Texas Advanced Computing Center where he works on the Agave “science as a service” project, a hosted platform for hybrid cloud, HPC and high-throughput scientific computing. Joe is co-creator of several open source Python projects in use at TACC including abaco, a system that implements the actor model of concurrent programming using containers and http.
| Attachment | Size |
|---|---|
 SEA2016_Polyglot.pdf | 2.5 MB |
