*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Control plane for situation-awareness applications in geo-distributed resources
Enrique Saurez
School of Computer Science
College of Computing
Georgia Institute of Technology
https://www.cc.gatech.edu/~esaureza/
Date: Monday, March 8th, 2021
Time: 3:00 PM - 5:00 PM (EST)
Location: https://bluejeans.com/316756759
Committee:
Dr. Kishore Ramachandran (Advisor, School of Computer Science, Georgia Institute of Technology)
Dr. Mostafa Ammar (School of Computer Science, Georgia Institute of Technology)
Dr. Alexandros Daglis (School of Computer Science, Georgia Institute of Technology)
Dr. Ada Gavrilovska (School of Computer Science, Georgia Institute of Technology)
Dr. Bharath Balasubramanian (Principal Scientist, AT&T Labs Research )
Abstract:
Situation-awareness applications generate actionable knowledge from sensor and user data. Two trends are unlocking new situation-awareness applications: geo-distributed computational resources and pervasive sensors. Geo-distributed computing infrastructure is now available worldwide, ranging from multi-region cloud deployments to newer 5G deployments at the last mile of the network. Pervasive sensors utilizing this infrastructure have also seen a boom, with examples like geo-distributed camera deployments, smart cities, and users' gadgets (e.g., smartphones) becoming increasingly common in everyday life. Having computational resources closer to the data generators improves both the response time and the efficiency in using the available resources. However, the efficient management of resources is challenging for densely geo-distributed resources because of the increased latency between components (due to physical distance) and the comparatively reduced resources in each location (due to space constraints at the last mile).
The first part of this thesis proposes a flexible control plane architecture. The control plane for geo-distributed resources has an inherent trade-off between response time and quality of responses. The main factor that defines the trade-off is the transmission latency that needs to be incurred between the resources being handled and the control plane component making those decisions for the resources. For example, if the control plane decisions are made in a centralized fashion for all the resources, we can calculate optimal decisions, but there is a high likelihood that all these decisions would incur at least one wide-area network round-trip given that the resources are geo-distributed. We present an architecture that defines the different building blocks required to manage geo-distributed resources efficiently. These building blocks are flexible and can be configured to achieve different trade-offs.
The second section of this thesis explores the two components in the architecture that are more affected by higher inter-component communication latency: state management and scheduling. The current cloud datacenter solutions assume low-latency communication across distributed components and incur heavy penalties when this assumption is violated. For state management, we analyze how we can optimize the architecture for the state of geo-distributed resources. We present mechanisms that reduce the required coordination round-trips while maintaining the same guarantees and semantics as cloud solutions by optimizing for the application structure. Similarly, we present scheduling optimizations to use the available resources better and reduce the number of communication messages required. The scheduling optimizations leverage application semantics and the computational resource locations to improve the distribution of control plane decisions among the architecture components.
Additional Meeting Details:
Meeting URL
https://bluejeans.com/316756759?src=join_info
Meeting ID
316 756 75
Want to dial in from a phone?
Dial one of the following numbers:
+1.408.419.1715 (United States(San Jose))
+1.408.915.6290 (United States(San Jose))
(see all numbers - https://www.bluejeans.com/numbers)
Enter the meeting ID and passcode followed by #
Connecting from a room system?
Dial: bjn.vc or 199.48.152.152 and enter your meeting ID & passcode
