DSP: Dynamic Service Placement with Reconfiguration Cost and Delay

This paper proposes a dynamic service placement policy that maximizes the benefits of a service provider with a limited average investment budget taking into account the reconfiguration delay and cost. The proposed algorithm captures (i) a tradeoff between the backhaul bandwidth usage and the lease of edge computing resources for services, (ii) a management of the spatio-temporal budget investment and (iii) a tradeoff between obtaining the benefit by placing new services at the edge servers and sticking to current states without the reconfiguration. Finally, we verify the performance of the proposed policy via simulations.