Programming Abstractions for Data Stream Processing Systems
Speaker: Konstantinos Mamouras, University of Pennsylvania
Location: 60 Fifth Avenue 150
Date: February 23, 2018, 11 a.m.
Host: Thomas Wies
Modern information processing systems increasingly demand the ability to continuously process incoming streaming data in a timely and reliable manner. Data streams arise in diverse applications ranging from patient monitoring in healthcare to real-time decision-making in emerging Internet of Things (IoT) systems. In this talk, I will present my research on the design of programming abstractions for stream processing that enable guarantees of correctness and predictable performance. First, I will present StreamQRE, a declarative domain-specific language and execution engine for stream processing. StreamQRE offers strong theoretical guarantees for resource usage, and its performance on realistic workloads is shown to compare favorably against other popular streaming engines. As a case study, I will discuss the application of StreamQRE to the design space exploration of alternative algorithms for cardiac arrhythmia detection. Finally, I will introduce a type-based framework for the logical specification of distributed streaming computations that facilitates correct and efficient deployment on distributed architectures such as Apache Storm.
Konstantinos Mamouras is a postdoctoral researcher at the University of Pennsylvania in the Department of Computer and Information Science. He is interested in the application of programming language techniques to the design of reliable and efficient software systems. He obtained his Ph.D. from Cornell University in 2015 under the supervision of Professor Dexter Kozen, where he worked on program semantics and logics for verification.
Refreshments will be offered starting 15 minutes prior to the scheduled start of the talk.