In cloud environment dynamic resource allocation is a problematic situation. To resolve this problem, the authors propose a gossip protocol that ensures fair resource allocation on their sites and applications, dynamically scales and adapts both number of physical machines and allocation to the load changes. They present a protocol that computes an optimal solution without considering memory constraints and prove correctness and convergence properties. This gossip protocol provides an heuristic solution for minimizing the cost of allocation. This paper will ensure continuous execution and don't require global synchronization.