Adaptive and Incremental-Clustering Anomaly Detection Algorithm for VMs Under Cloud Platform Runtime Environment

Abstract

The advent of cloud platform has promoted the complexity and scales of industries increasingly. Any deliberate or non-deliberate faults may cause enormous impact on system performance and server costs. Anomaly detection is a good way to identify anomalies and improve the dependability of the cloud platform. However, some of the anomaly detection methods are labeled data dependency, and some of them are sensitive to the dynamic runtime environment of the cloud platform. To address the problems, an adaptive and incremental clustering anomaly detection algorithm for virtual machines under the cloud platform runtime environment is proposed. Compared with the previous detection methods, the effect of the runtime environment factor is taken into account. Owning to the high level of dynamic cloud platform manages and the resources allocation of virtual machines, the environmental factors play an important role in the running performance of the virtual machines. In this paper, an improved adaptive and incremental clustering algorithm is introduced to perform the detection with the considerations of the cloud platform runtime environment. To demonstrate the effectiveness, two sets of experiments are performed. The experimental results indicate that the proposed anomaly detection method can greatly improve the detection accuracy rate even the cloud platform runtime environment changes.

Keywords

• Anomaly detection
• runtime environment
• cloud platform
• incremental clustering
• virtual machine