Multi-Layered Attack Recognition (MLAR) Model to Protect Cloud From EDOS Attacks

In this paper, the work is carried upon the security of cloud computing environment from the economic denial of service (EDOS) attacks. The cloud computing environments have become popular in the past decade, as number of businesses has shifted their work online to the cloud. The cloud service providers offer the maximum uptime for their users, which is committed in the form of service level agreements (SLAs). The uptime is the term used to indicate the availability of the cloud computing resources for client's application to run its processes. If at any point, the client application is not able to run its operations due to non-availability of resources, it is considered as the downtime, which is negative to the uptime and considered as SLA violation. The SLAs keep the information about the SLA violations and their settlements, which is considered as the loss of the cloud service providers. For example, the uptime of 99.99% is committed by cloud service provider, and due to any internal or external reason, achieved uptime is lower than committed value, the cloud service provider may face a penalty as per defined in the SLA, which is considered direct profit loss for the service provider. In order to reduce the financial losses due to the EDOS attacks, the proposed model is designed by combining the periodic authentication, pattern analysis and data flow control mechanisms to prevent the cloud from attacks. The proposed Multi-Layered Attack Recognition (MLAR) Model has outperformed the existing controlled access based EDOS (CA-EDOS) prevention model on the basis of resource utilization and response delay parameters.