Provisioning services for applications in networks of large data-centers is carried out as service function chaining (SFC) by linking ordered list of functions. With the development of software defined network (SDN) and network function virtualization (NFV), SFC has gained more attention in the industry for deploying larger applications. Provisioning different virtualized functions (VFs) of an SFC in different data-centers in a multi data-center environment is a challenging task. Existing works on serving SFC select a service function path (SFC) along which the VFs are executed. However, none of the works provide the optimal solutions considering both the execution and data transfer delay along with optic electric optic (OEO) conversion delay. In this work, a global optimal solution for SFP selection has been developed considering queuing, processing, transmission and OEO conversion delay. The proposed dynamic solution can work in a heterogeneous environment. Performance evaluation of the proposed algorithm shows that, significant performance improvement in respect of total execution delay of an SFC can be achieved compared to state-of-the-art works.