Abstract The load balancing (LB) may be used at different levels to reduce overhead for the decision‐making process. In the past decade, cloud computing has drawn a lot of attention from both the academic and commercial communities to get demanded resources (machines, platforms, data, storage, software, and so forth) as a service on rent economically. Generally, a situation may arise when requests are not meeting their deadlines and the cloud provider wants to finish the running application in minimum time. In this article, a receiver initiated deadline aware LB strategy (RDLBS2) has been proposed which attempts the migration of incoming cloudlets to appropriate virtual machines (VMs) where the deadlines of the cloudlets are met to optimize the turnaround time by exploiting the remaining processing capacities of VMs. A simulation study has been carried out by using Cloud‐Sim as a simulator. A sensitivity analysis has been presented to analyze the effects on performance parameters by varying the number of cloudlets and the number of VMs while keeping the remaining input parameters fixed. The experimental evaluation and analysis suggest that RDLBS2 performs significantly better than its peers on objective parameters almost in all cases under study.