Non-Stationary Gradient Descent for Optimal Auto-Scaling in Serverless Platforms

TL;DR

This paper introduces a stochastic gradient descent method for optimal auto-scaling in serverless platforms, addressing non-stationary conditions and providing convergence guarantees, with validation through simulations.

Contribution

It develops a novel non-stationary stochastic optimization approach for auto-scaling, with proven asymptotic optimality and convergence rate, tailored for transient system behaviors.

Findings

The proposed method achieves asymptotic optimality almost surely.

Convergence rate of the scheme is $O(n^{-2/3})$.

Numerical simulations show improved performance over existing rules.

Abstract

To efficiently manage serverless computing platforms, a key aspect is the auto-scaling of services, i.e., the set of computational resources allocated to a service adapts over time as a function of the traffic demand. The objective is to find a compromise between user-perceived performance and energy consumption. In this paper, we consider the \emph{scale-per-request} auto-scaling pattern and investigate how many function instances (or servers) should be spawned each time an \emph{unfortunate} job arrives, i.e., a job that finds all servers busy upon its arrival. We address this problem by following a stochastic optimization approach: we develop a stochastic gradient descent scheme of the Kiefer--Wolfowitz type that applies \emph{over a single run of the state evolution}. At each iteration, the proposed scheme computes an estimate of the number of servers to spawn each time an unfortunate job arrives to minimize some cost function. Under natural assumptions, we show that the sequence of estimates produced by our scheme is asymptotically optimal almost surely. In addition, we prove that its convergence rate is $O(n^{-2/3})$ where $n$ is the number of iterations. From a mathematical point of view, the stochastic optimization framework induced by auto-scaling exhibits non-standard aspects that we approach from a general point of view. We consider the setting where a controller can only get samples of the \emph{transient} -- rather than stationary -- behavior of the underlying stochastic system. To handle this difficulty, we develop arguments that exploit properties of the mixing time of the underlying Markov chain. By means of numerical simulations, we validate the proposed approach and quantify its gain with respect to common existing scale-up rules.