Server-features performance evaluation of node in p2p networks

Performance improvement is the fundamental goal of hardware upgrades, system tuning, and proposing and implementing novel algorithms. Apparently, how to quickly and accurately evaluate system performance is one of the most critical steps in this process. Usually, performance evaluations are performed on computers with specific software and hardware features. Therefore, the simple run-and-test model lacks platform and software independence, which makes the results far from being objective and less convincing despite the fact that those applications are platform independent. In this paper, we focus on peer-to-peer (p2p) systems and propose a novel performance evaluation where a series of carefully constructed tests are conducted on two different platforms. The striking difference between our model and traditional model is that instead of treating one p2p node as a single identity, we split the client features and server features of each p2p node and theoretically build the test sets. We argue that the model we proposed can more accurately reflect the system performance in a real p2p system