Online Learning for Function Placement in Serverless Computing

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Huang, Wei, Combes, Richard, Araldo, Andrea, Castel-Taleb, Hind, Jouaber, Badii
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908390049972224
author Huang, Wei
Combes, Richard
Araldo, Andrea
Castel-Taleb, Hind
Jouaber, Badii
author_facet Huang, Wei
Combes, Richard
Araldo, Andrea
Castel-Taleb, Hind
Jouaber, Badii
contents We study the placement of virtual functions aimed at minimizing the cost. We propose a novel algorithm, using ideas based on multi-armed bandits. We prove that these algorithms learn the optimal placement policy rapidly, and their regret grows at a rate at most $O( N M \sqrt{T\ln T} )$ while respecting the feasibility constraints with high probability, where $T$ is total time slots, $M$ is the number of classes of function and $N$ is the number of computation nodes. We show through numerical experiments that the proposed algorithm both has good practical performance and modest computational complexity. We propose an acceleration technique that allows the algorithm to achieve good performance also in large networks where computational power is limited. Our experiments are fully reproducible, and the code is publicly available.
format Preprint
id arxiv_https___arxiv_org_abs_2410_13696
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Online Learning for Function Placement in Serverless Computing
Huang, Wei
Combes, Richard
Araldo, Andrea
Castel-Taleb, Hind
Jouaber, Badii
Machine Learning
Networking and Internet Architecture
We study the placement of virtual functions aimed at minimizing the cost. We propose a novel algorithm, using ideas based on multi-armed bandits. We prove that these algorithms learn the optimal placement policy rapidly, and their regret grows at a rate at most $O( N M \sqrt{T\ln T} )$ while respecting the feasibility constraints with high probability, where $T$ is total time slots, $M$ is the number of classes of function and $N$ is the number of computation nodes. We show through numerical experiments that the proposed algorithm both has good practical performance and modest computational complexity. We propose an acceleration technique that allows the algorithm to achieve good performance also in large networks where computational power is limited. Our experiments are fully reproducible, and the code is publicly available.
title Online Learning for Function Placement in Serverless Computing
topic Machine Learning
Networking and Internet Architecture
url https://arxiv.org/abs/2410.13696