Performance Analysis Of Short Message System Based On Performance Model

One of important software quality characters is performance. Software's performance always is associated with its architecture closely. And it seems more obviously to large-scale distributed system. System performance bottlenecks which caused by unsuitable system architecture design hard to discover until the last system integration test phase. Any changing of system architecture would be very expensive at that moment. It needs the ability of finding that design flaws out and correcting them in design phase. The solve method that analyze Performance model which abstracted from UML model that build in system design phase is widely accepted at present. It belongs the software performance engineering (SPE)'s research field.On the other hand, since the rapid development of short message business, the performance problem of short message system (SMS) is coming up. Building high-performance SMS becomes the focus of every telecom equipment support and telecom serve provider. The key point of that is how to configure a SMS to satisfy system's performance requirement.QN performance model abstracted from UML model of SMS in this paper. Using APERA performance model solve tool to analyze QN model, this paper present a systematic method to find the configuration of SMS, which can satisfy performance requirement. The method also suits for other large-scale systems. It can import performance analysis in system design phase to reduce design flaws. And in system deployment phase it can guarantee system satisfied performance requirement at the lowest hardware cost.How to improve the performance of SMS at high intensity workload is another research topic of this paper. By develop short message performance test Demo and analyze its performance, this paper presents present a threads dynamic configuration policy which can make SMS gain the ability to balance submit and deliver workload and get the max system throughput at the same time. It can availably avoid that system throughput descend at high intensity workload. And specially, to short message service center (SMSC), this paper present a new workload-based resend policy of SMSC for workload adaptability. Using the resend policy, SMSC can control its deliver workload availably; reduce the risk of SMSC congestion at high intensity workload.