A Collaborative Load Balancer for Network Intrusion Detection in Cloud Environments

Abstract

Information security and permeability in the system is a major concern for cloud computing. Cloud service providers should ensure that user information remains private from other people (external or internal). Deployment of an intrusion detection system (IDS) is a technique to protect the cloud from existing security and intrusion threats. Due to the high traffic volume and the need for flexibility and the diversity of applications in a cloud environment, a distributed IDS is used to apply traffic management, and provide scalability concurrently. The challenge is to propose a method of load balancing that distributes input traffic among IDS sensors, and optimally balance the workload on the sensors, to improve the overall performance of the IDS, with minimum degradation in quality. In this paper, a load balancing method is presented which operates based on the sensors' hardware specification and their compatibility with incoming requests, and the sensors' available resources. Dynamic request allocation to sensors is done in real time, using application layer load distribution, with no need to migrate requests or sensors, which could lead to extra overhead. The accuracy of IDS sensors in detection of attacks can be affected by the distribution of input traffic, but this shortcoming is resolved by our method. We compare the proposed method with existing algorithms in terms of load balancing and IDS functionality. The results confirm superior performance of IDS functionality in the proposed architecture.