This is a 266 page report. University of California Santa...

This is a 266 page report.
University of California Santa Barbara
Autonomous and Predictive Systems to Enhance the Performance of Resilient Networks.
A dissertation submitted in partial satisfaction of the requirements for the degree
Committee in charge:
Professor Elizabeth Belding, Chair Professor Arpit Gupta
Professor Trinabh Gupta
July 2022
Doctor of Philosophy in Computer Science
by Vivek Adarsh
Our study proposes AKIDA, a new architecture that strategically harvests the
untapped compute capacity of the SmartNICs to offload transient workload 10
spikes, thereby reducing the SLA violations. Usage of this untapped compute capacity is more favorable than adding and deploying additional servers, as SmartNICs are economically and operationally more desirable. AKIDA is a low-cost and scalable platform that orchestrates seamless offloading of server- less workloads to the SmartNICs at the network edge, eliminating the need for pre-allocating expensive compute power and over-utilization of host servers. Our system evaluation shows