Bilkent University
Department of Computer Engineering
S E M I N A R

 

A co-routine based auto-parallel stream processing engine

 

Semih ahin

Stream processing is a computational paradigm that enables processing of analytical tasks in an efficient and scalable manner. In this paradigm, incoming data stream is processed on-the-fly. Streaming applications are structured as data flow graphs where vertices represent operators and edges represents data streams. In stream processing, task and pipeline parallelism are exploited by nature. In this research, we developed co-routine based stream processing engine. In addition to task and pipeline parallelism, we exploit data parallelism with our auto-parallelism module. We have also proposed different scheduling algorithms for different purposes such as high throughput or low latency.

 

DATE: 08 December, 2014, Monday @ 16:50
PLACE: EA-409