Managing Configuration Errors in Cloud Computing Using Cluster & Node Management Techniques

Abstract

Cloud computing has been the technology of choice for most users globally due to the numerous advantages that it offers. This technology is offered to users with different levels of expertise who utilize the different services available in the cloud. At infrastructure level, however, configuration errors are some of the major causes of unavailability of services to users. The objective of this study was to evaluate whether cluster management and node management could effectively be used to manage configuration errors proactively at cloud infrastructure level. The literature indicates that most configuration errors occur due to human error, namely parametric configuration errors. Some human errors at infrastructure level were simulated using CloudSim simulation toolkit. The availability of the infrastructure was then measured using two availability parameters, namely, service availability and execution availability. The simulations showed that the injected errors resulted in a partial or complete loss of services to users. The correct configurations were then simulated to show the difference, and suggestions are made as to how these errors may be prevented proactively in the first place. The findings showed that cluster management is an effective tool in preventing configuration errors as long as those errors occur at cluster level; further node management is an effective tool in countering configuration errors, as long as the errors occurred at node level. It would thus be more desirable to use cluster management techniques to counter errors at both node and cluster level as the latter management technique presents management at a more abstract level