A secure VM allocation scheme to preserve against co-resident threat

Abstract

Preserving the secrecy in cloud system is one of the biggest concerns for the cloud customers who faces security risks in the context of load balancing. The co-resident attacks are widely used by attackers, where malicious users build side channels and extract private information from VMs. The proposed model evaluates the possibility of VM co-residency and success rate of an attack. The emphasis of this paper is to reduce the possibility of co-resident attacks among different users. When cloud data centres receive requests for the tasks deployment, then the proposed system will find out the secure physical machine under VM allocation policies while avoiding the threats. The performance is calculated by these metrics: makespan, resource utilisation, co-residency probability and co-resident success rate. The results show that the most virtual machine allocation policy (MVMP) effectively reduces the risk under the safe states. The framework significantly improves the security by reducing the shared servers up to 32.2% and enhances the resource utilisation up to 44.14% over least VM allocation policy (LVMP), round robin VM allocation policy (RRVMP) allocation schemes.