Reducing Noisy-Neighbor Impact with a Fuzzy Affinity-Aware Scheduler

Abstract

Abstract-Overbooking techniques have been proven efficient to increase overall utilization of cloud datacenters. However, overbooking may also degrade applications performance as (at least) some applications need to share physical resources such as CPU or memory. Consequently, interference may increase among the virtual machines that share resources, the so called noisy neighbors effect. We present an affinity-aware scheduler to reduce the impact of such interference. A fuzzy logic engine accounts for the uncertainty in these environments and estimates which CPU cores are currently more suitable for each incoming application. This helps the scheduler make virtual machine to physical resource mapping decisions, also known as vcpu pinning. An experimental evaluation based on a combination of interactive services and batch applications confirms that our affinity-aware fuzzy scheduler reduces the interference among applications, enabling more predictable performance and consequently safer overbooking.

Using a Digital Twin as the Objective Function for Evolutionary Algorithm Applications in Large Scale Industrial Processes

“…An important aspect in cloud computing is the noisy neighbour problem, described in greater detail by [68] and [69]. The servers typically used in cloud computing are virtual machines, hosted on a physical server with multiple other virtual machines.…”

Impact of etcd deployment on Kubernetes, Istio, and application performance

“…Cloud infrastructure providers have monitoring tools that provide both the required insight into resource usage and the administrative powers required to perform high-level optimization of the infrastructure. [32][33][34][35] Although we used a private cloud in this work, we were neither the sole users (ie, we had no dedicated resources) nor did we have the administrative powers required for such mitigation techniques. Cloud users have limited ability to detect and mitigate the effects of noisy neighbors because they lack access to the relevant data and have insufficient permissions to affect the deployment.…”

Scalable Prediction of Service-Level Events in Datacenter Infrastructure Using Deep Neural Networks

“…Several works have modelled VMs placement as a multi-dimensional optimization problem that can be solved applying specific heuristics [8], [9]. Complementary, a work based on the application of fuzzy logic for safe VM placement and noisy neighbor avoidance was recently proposed in [7]. In [12] authors propose to leave some datacenter capacity unused in the underlying infrastructure in order to provide with free resources to the applications that start suffering the noisy neighbor effect.…”