Distributed Cloud Brokerage: Solution to Real World Service Provisioning Problems

This research analyzes the performance of a distributed cloud broker in a live cloud environment, utilizing a government owned, private, federated cloud. The paper explores functioning of a distributed cloud broker that assists in the provisioning of services to geographically distributed data centers. The data centers have volunteered to federate and expose their utilization metrics to each other through the cloud broker. The experimental infrastructure utilizes the closed schema for federation. The cloud broker is responsible for match making and bundling/provisioning of services from multiple private cloud providers, through volunteer and federated data centers. This has been tested under multiple load conditions. The proposed distributed cloud broker handles the load on the cloud ecosystem through a strictly controlled mechanism in a private cloud ecosystem, custom routing all overload conditions (cloud bursting scenario) on specific private clouds through a common interface visible to the Amazon Web Services (AWS). The proposed broker mechanism shows high efficiency and lesser cloud bursting instances compared to a pure AWS based ecosystem. The work also analyzes real world issues faced by organizations handling cloud brokerage frameworks in a distributed manner. The research asserts that it is possible to create customized distributed cloud brokers and the perspective of using a hybrid cloud approach using distributed broker in federated clouds is feasible, albeit in a tightly integrated and fine tuned cloud environment.