An Elastic Layers Pattern Approach with Dynamically Added Layers

Abstract

Cloud environments often provide dynamic resource allocation techniques. These can be used to scale single components or even whole software systems according to their current workload. Fluctuating workloads can occur in different layers of the software architecture and need appropriate handling to meet performance requirements. Scaling workload-intensive components in combination with load-balancing can be used to deal with these issues. Therefore, we present a parallel layers approach, which extends an existing pattern regarding improved elasticity. Based on dynamically adding layers on top of bottleneck layers, we increase the flexibility and performance of related architectures. Furthermore, we describe a first design approach, implementation and an evaluation of the feasibility.