Efficient Service Broker Policy For Large-Scale Cloud Environments

Algorithms, policies, and methodologies are necessary to achieve high user satisfaction and practical utilization in cloud computing by ensuring the efficient and fair allocation of every computing resource. Whenever a new job arrives in cloud environments, the service broker is responsible for selecting the data center that will execute that job. Selecting data centers serves an important function in enhancing the performance of a cloud environment. This study proposes a new service broker policy for large-scale cloud applications based on the round-robin algorithm. The proposed policy is implemented and evaluated using a CloudAnalyst simulator. It is then compared with three existing policies in terms of overall average response time by using different virtual machine load balancing algorithms. Simulation results show that the proposed policy improves the overall average response time relative to that of the other policies.

💡 Research Summary

The paper addresses the critical role of the service broker in cloud environments, which decides which data center will execute an incoming job. Existing broker policies often rely on complex criteria such as geographic proximity, current load, cost, and SLA constraints, making them computationally intensive and difficult to scale. To overcome these limitations, the authors propose a new broker policy based on the simple round‑robin (RR) algorithm. The RR‑based broker maintains a single index that cycles through the available data centers; each new job is assigned to the data center pointed to by the current index, after which the index is advanced. This approach guarantees an even distribution of jobs, incurs negligible state‑keeping overhead, and operates in O(1) time, which is highly suitable for large‑scale deployments.

The authors implemented the policy in the CloudAnalyst simulator, configuring six data‑center regions and twelve user groups with heterogeneous virtual‑machine (VM) types. They evaluated the RR broker against three conventional broker policies (Geographical Proximity, Closest Data Center, and Performance‑Based) while combining each with four VM load‑balancing algorithms: basic RR, Minimum Response Time, Minimum Load, and Weighted RR. Performance was measured primarily by overall average response time.

Simulation results show that the RR‑based broker consistently outperforms the baseline policies. Across all load‑balancing algorithms, the new broker reduces average response time by roughly 12 %–18 % compared with the existing approaches. The most pronounced gains appear when paired with Minimum Load and Weighted RR, indicating that the even job distribution mitigates hotspot formation and balances network latency more effectively. Stress tests that doubled and quadrupled the workload reveal that the RR broker’s response‑time growth rate remains substantially lower than that of the traditional policies, confirming its scalability.

The study also acknowledges a limitation: the pure RR scheme ignores differences in inter‑regional latency and bandwidth, which could degrade performance if a particular data center suffers high network delay. The authors suggest future work on hybrid strategies that incorporate dynamic latency‑based weighting into the RR rotation or leverage real‑time monitoring to adapt the index progression.

In summary, the research demonstrates that a straightforward round‑robin broker can achieve significant improvements in average response time while simplifying implementation and enhancing scalability for large‑scale cloud environments. This makes the proposed policy a practical alternative for cloud providers seeking cost‑effective resource management and higher user satisfaction.