In this paper, analytical models are presented for modeling cloud load balancing mechanism combing queueing networks (QNs) and generalized stochastic Petri nets (GSPNs). To this end, a hybrid model is proposed to model a cluster inside an Infrastructure-as-a-Service (IaaS) cloud data center in the first step. The model includes several real aspects of such environments, such as request queueing, virtual machine (VM) provisioning, VM servicing, and powering on and off of physical machines (PMs). Based on the presented cluster model, a hybrid monolithic model is proposed in the second step that encompasses requests arrival and load balancing component of the cloud. The proposed monolithic model supports heterogeneity of requests in terms of the arrival process and the number and execution time of requested VMs. To demonstrate the applicability of the hybrid monolithic model, several load balancing algorithms that can be applied to such a model are introduced and evaluated based on different performance metrics of interest.