Current researches on SaaS(software as a service) optimization placement mostly assume that the types and number of virtual machines are constant in cloud environment, namely, the optimization placement is based on the restricted resource. However, in actual situation the types and number of virtual machines are unknown, and they need to been calculated according to the resource requirement of components deployed. To address the issue, from the view of SaaS providers, this paper proposes a new approach to SaaS optimization placement problem that not only is applied to initial deployment of SaaS, but also is applied to component dynamic deployment in the running phase of SaaS. A hybrid genetic and simulated annealing algorithm(HGSA) is used in this approach that combines the advantages of genetic algorithm and simulated annealing algorithm, and overcomes the problems of the premature of genetic algorithm and the lower convergence speed. Compared with the separated using of genetic algorithm and simulated annealing algorithm, the experimental results show that HGSA has higher quality in solving the problem of SaaS component optimization placements. The approach proposed in this paper will provide the support of theory and method for the large-scale application of SaaS service mode.