This document proposes an efficient and reliable resource management framework for public cloud computing. It consists of a gossip protocol that ensures fair resource allocation among sites by calculating memory and CPU load factors. It also includes a routing table for dynamically managing tasks. A request partitioning approach is proposed based on the gossip protocol to facilitate cost-efficient splitting of user requests among cloud service providers. Following request partitioning, resource embedding is performed to map virtual to physical resources efficiently and balance resource allocation. The framework is evaluated on a simulated cloud environment and is shown to provide reliable and dynamic resource management.