Events Calendar

PhD Defense - Alok Gautam Kumbhare

**Ph.D. Defense Announcement**

"Adaptive and Resilient Stream Processing on Clouds"

Ph.D. candidate: Alok Gautam Kumbhare

Wednesday, January 13, 2016
10:00AM -“ 12:00PM
EEB 248


Abstract:
Ubiquitous deployment of physical and virtual sensors, coupled with tremendous increase in the number of connected devices has led to the explosion of data, not only in terms of the volume and but also the velocity at which it is being generated. Scalable stream processing systems are necessary to process these high velocity data streams and to derive useful insights in real-time. However, unlike high volume batch processing, stream processing applications need to run continuously with minimum downtime and hence are susceptible to the three dimensions of dynamism: domain dynamism -“ variations in domain requirements; infrastructure dynamism -“ temporal and spatial variations in infrastructure performance and system failures; and data dynamism -“ changes in input data rates over time, all of which adversely affect the application QoS and value achieved from the application.
We posit that applications that are aware of this dynamism and can adapt to the changing conditions are critical in achieving the desired QoS in a cost-efficient manner. We thus propose a novel Dynamic Dataflow Application model and execution framework that inherently supports flexible stream processing applications and allow seamless runtime recomposition. The model provides an extremely powerful tool to develop, deploy and execute long-running dynamic applications with minimum overhead and promotes the notion of value-driven execution through continuous adaptations to achieve the best value from the application.
We also propose several scheduling and resource mapping heuristics for deployment of these dynamic dataflows on public clouds that take advantage of the pay-as-you-go cost model to achieve the desired quality of service (QoS) and provide a balance between the application value, and resource cost. Finally, to ensure uninterrupted execution in the presence of infrastructure failures, we propose novel integrated approach for efficient runtime elasticity, fault-tolerance and load balancing that enables seamless scaling and provide high fault-tolerance with sub-second recovery latency and hence offer strong resilience and service guarantees.

Bio:
Alok Gautam Kumbhare is a PhD candidate at the department of Computer Science, University of Southern California under the guidance of Prof. Viktor K. Prasanna and Prof. Yogesh Simmhan (IISc Bangalore). His primary research focus is on developing adaptive scheduling algorithms and large-scale distributed systems for processing high-velocity and highly variable data streams on Cloud computing environments. His research interests also include big data analytics and large scale machine learning, resource management, and graph analytics. He completed his B.Tech from Indian Institute of Technology Guwahati in 2008.


Defense Committee: Prof. Aiichiro Nakano, Prof. Viktor K. Prasanna (chair), Prof. Cauligi Raghavendra

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Lizsl De Leon