Cloud Brokering Architecture for Dynamic Placement of Virtual Machines

Abstract

Numerous service providers in the cloud market provide diverse services, at distinct price and level of competence. However, the concept of cloud computing is yet to reach many organizations and to the end users as well. With perspective of these organizations and end users, it is very difficult to choose one among many service providers. Further, adhering to only one provider will deprive these users of benefits from market diversity and fluctuating rates. Instead, having a broker be aware of the provider's service offerings and consumer's requirements, and managing the assignment under economic and technical constraints will benefit both ends of the service viz consumers and providers. In this work, a cloud resource broker is proposed that will govern the assignment of providers' resources to consumer dynamically. The proposed broker uses various requirements and constraints specified by the consumer in the requirement description template as input, to calculate aggregated requirements, using an aggregation algorithm. Further, the service scheduling algorithm is defined to find an optimized match between the aggregated requirements with the providers offerings. Thereafter, this algorithm is executed frequently, based on a strategy for dynamic scheduling to benefit consumers on account of introduction of new provider or some good offerings. Results show that the solution provided by broker proves to be a win-win situation for the consumer with respect to cost as well as performance.