Back To Schedule
Thursday, November 16 • 9:50am - 10:30am
Cloud Science

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.
Cloud Science
At GRAIL, our mission is to detect cancer early, when it can be cured. Our approach is data intensive: we sequence cell-free DNA in blood in order to detect minuscule evidence of tumors.
In order to support our myriad workloads—among them ad-hoc analyses, model training, and complicated bioinformatics pipelines—we built Reflow, a system and language for scientific computing in the cloud. Reflow’s data processing engine is fully incremental, and focuses on efficiency, reproducibility, and ease-of-use.
With Reflow, scientists and engineers write ordinary programs that compose existing tools; these programs are then transparently parallelized, memoized, and distributed across many workers using your favorite cloud computing provider. Reflow is vertically integrated: a single binary evaluates the program and is also responsible for elastic cluster management and execution coordination — this makes Reflow very simple to deploy, operate, and retarget to different cloud providers.
I’ll describe how Reflow’s language semantics and runtime are co-designed to yield a simple, robust implementation. I’ll also talk about how Reflow is used for data intensive scientific computing at GRAIL, primarily to analyze next generation sequencing (NGS) data sets.

avatar for Marius Eriksen

Marius Eriksen

infrastructure-infrastructure engineer, Grail
Marius is the author of such ideas as Finagle, Zipkin, Your Server is a Function, and many others.

Thursday November 16, 2017 9:50am - 10:30am PST