Dynamic Resource Allocation and Load Balancing in Edge Computing Systems
||Dynamic Resource Allocation and Load Balancing in Edge Computing Systems|
||SNIC Small Compute|
||Feridun Tütüncüoglu <email@example.com>|
||Kungliga Tekniska högskolan|
||2021-02-17 – 2022-03-01|
Edge computing is expected to serve a variety of applications with stringent latency constraints. At the same time, it is expected to provide a lightweight programming abstraction and resource management. We consider the interplay between latency constrained applications and function-level resource management in edge computing. We develop a model of the interaction between applications that maximize their utility through rate adaptation and a load balancing operator under a function-oriented pay-as-you-go pricing model. We show that the strategic interaction between the applications can be formulated as a generalized Nash equilibrium problem, and use variational inequality theory to prove that the game admits an equilibrium. We show that simple rate adaptation may lead to starvation, and we propose an online algorithm for applications to maximize their utility through rate adaptation and resource reservation. We show that the algorithm achieves zero regret, and our simulation results show that the algorithm achieves good system performance at equilibrium, ensures fast convergence, and allows applications to meet their latency constraints.