Hierarchical Service Placement for Demanding Applications
Elisa Maini, Truong Khoa Phan, David P. Griffin, Miguel Rio · 2016
The increasing scale of cloud environments requires more scalable orchestration systems for determining which physical resources are responsible for processing service requests. A centralised service placement lacks scalability while a fully decentralised approach only has a limited view of the system. For these reasons, this paper investigates a hierarchical approach for service placement in a distributed environment, which increases scalability while maintains high service placement quality. First, we design a polynomial optimisation algorithm to place services in cloud data centers based on our novel utility function. Then, we describe a hierarchical model with the need to only know a small subset of the data required by the global optimisation formulation. Simulations show that our approach is scalable and performs well, close to the centralised model.