View article

The Internet of services, as a global service oriented architecture, is a complex large scale system that presents many problems in the design and implementation of systems to autonomously manage, protect and tune it. Whilst the emerging data centres or service `cloud' represent a move towards centralization, computing power is becoming disembodied and is consumed where and when it is needed in a decentralized manner. Thus this paper investigates the use of ambient type entities, to identify and optimize core structures as bounded service ambients, to address the gap between desired global outcome and service actions. The approach is assessed against an implemented decentralized approach to load balancing at a server farm.