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

 

Locality Aware State Management in Distributed Stream Processing Applications

 

Muhammed Yaðmur Þahin

Distributed stream processing is a way to process series of data, called streams, on-the-fly in a distributed manner. Scalability and high throughput is crucial for these types of applications. In order to satisfy the needs of such applications locality aware state partitioning and management mechanism can be utilized. Locality aware state partitioning provides distributing and storing relative data close to each other. In this research, we worked on a locality aware state partitioning and management mechanism in order to provide scalability and high throughput via reducing disc access. We proposed algorithms for distributing data according to its locality as well as a pluggable state management mechanism for Storm stream processing engine.

 

DATE: 08 December, 2014, Monday @ 16:15
PLACE: EA-409