Handling Flash-Crowd Events to Improve the Performance of Web Applications

TL;DR

This paper introduces a formal model and a novel algorithm, FCHP-ILS, for efficiently handling flash-crowd events in web applications using cloud resources, optimizing resource allocation and content replication.

Contribution

It formulates the Flash Crowd Handling Problem as an integer programming model and proposes the FCHP-ILS algorithm for improved resource management during flash crowds.

Findings

FCHP-ILS effectively handles flash crowds with better resource utilization.

The approach outperforms Amazon's Auto-Scale in case studies.

Real trace-based evaluation demonstrates its practical benefits.

Abstract

Cloud computing can offer a set of computing resources according to users' demand. It is suitable to be used to handle flash-crowd events in Web applications due to its elasticity and on-demand characteristics. Thus, when Web applications need more computing or storage capacity, they just instantiate new resources. However, providers have to estimate the amount of resources to instantiate to handle with the flash-crowd event. This estimation is far from trivial since each cloud environment provides several kinds of heterogeneous resources, each one with its own characteristics such as bandwidth, CPU, memory and financial cost. In this paper, the Flash Crowd Handling Problem (FCHP) is precisely defined and formulated as an integer programming problem. A new algorithm for handling with a flash crowd named FCHP-ILS is also proposed. With FCHP-ILS the Web applications can replicate contents in the already instantiated resources and define the types and amount of resources to instantiate in the cloud during a flash crowd. Our approach is evaluated considering real flash crowd traces obtained from the related literature. We also present a case study, based on a synthetic dataset representing flash-crowd events in small scenarios aiming at the comparison of the proposed approach against Amazon's Auto-Scale mechanism.