Performance optimization in cloud computing is of vital importance to ensure efficient resource use and to meet the demands of users. In this paper, we propose a framework that will optimize the efficiency of the cloud system through connection pooling. Traditional ways are normally inefficient due to continuous connection creation and termination, which adds to the overhead and causes delay. Our proposed framework resolves these issues by integrating the connections and reusing them to decrease latency, increase throughput, and improve scalability. Simulations run under different load conditions, such as Low, Medium, and High, showed dramatic improvements in performance. For example, at Low Load, it had an average of 0.002579 seconds in query time, and the throughput became 3.33 transactions per second. At High Load, it gave an average of 0.005134 seconds in query time and a throughput of 5 TPS. These results prove that connection pooling does improve performance bottlenecks, allowing cloud infrastructure to handle most diversities of workloads. Conclusively, this framework is a hands-on approach to the optimization of cloud performance and will make any web application scale better and be responsive to growing online service demands.