Distributed Load Balancing with Workload-Dependent Service Rates

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Wenxin, Balseiro, Santiago R., Kleinberg, Robert, Mirrokni, Vahab, Sivan, Balasubramanian, Wydrowski, Bartek
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915407082815488
author Zhang, Wenxin
Balseiro, Santiago R.
Kleinberg, Robert
Mirrokni, Vahab
Sivan, Balasubramanian
Wydrowski, Bartek
author_facet Zhang, Wenxin
Balseiro, Santiago R.
Kleinberg, Robert
Mirrokni, Vahab
Sivan, Balasubramanian
Wydrowski, Bartek
contents We study distributed load balancing in bipartite queueing systems where frontends route jobs to heterogeneous backends with workload-dependent service rates. The system's connectivity -- governed by compatibility constraints such as data residency or resource requirements -- is represented by an arbitrary bipartite graph. Each frontend operates independently without communication with other frontends, and the goal is to minimize the expected average latency of all jobs. We propose a closed-loop policy called the Greatest Marginal Service Rate (GMSR) policy that achieves effective coordination without requiring knowledge of arrival rates. In a discrete-time stochastic model, we show that the behavior of our routing policy converges (almost surely) to the behavior of a fluid model, in the limit as job sizes tend to zero and job arrival rates are scaled so that the expected total volume of jobs arriving per unit time remains fixed. Then, in the fluid regime, we demonstrate that the policy attains an $ε$-suboptimal solution in $O(δ+ \log{1/ε})$ time from $δ$-suboptimal initial workloads, which implies global convergence to the centrally coordinated optimal routing. Finally, we analyze the fluid model when the system is overloaded. We show that GMSR lexicographically maximizes throughput, maximizes the number of stable backends, and minimizes their collective workload.
format Preprint
id arxiv_https___arxiv_org_abs_2411_17103
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Distributed Load Balancing with Workload-Dependent Service Rates
Zhang, Wenxin
Balseiro, Santiago R.
Kleinberg, Robert
Mirrokni, Vahab
Sivan, Balasubramanian
Wydrowski, Bartek
Distributed, Parallel, and Cluster Computing
We study distributed load balancing in bipartite queueing systems where frontends route jobs to heterogeneous backends with workload-dependent service rates. The system's connectivity -- governed by compatibility constraints such as data residency or resource requirements -- is represented by an arbitrary bipartite graph. Each frontend operates independently without communication with other frontends, and the goal is to minimize the expected average latency of all jobs. We propose a closed-loop policy called the Greatest Marginal Service Rate (GMSR) policy that achieves effective coordination without requiring knowledge of arrival rates. In a discrete-time stochastic model, we show that the behavior of our routing policy converges (almost surely) to the behavior of a fluid model, in the limit as job sizes tend to zero and job arrival rates are scaled so that the expected total volume of jobs arriving per unit time remains fixed. Then, in the fluid regime, we demonstrate that the policy attains an $ε$-suboptimal solution in $O(δ+ \log{1/ε})$ time from $δ$-suboptimal initial workloads, which implies global convergence to the centrally coordinated optimal routing. Finally, we analyze the fluid model when the system is overloaded. We show that GMSR lexicographically maximizes throughput, maximizes the number of stable backends, and minimizes their collective workload.
title Distributed Load Balancing with Workload-Dependent Service Rates
topic Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2411.17103