Performance measurement results for some protocols using SPIMS

Different protocol performance measurements are difficult to compare as they are often performed with different protocol parameters, on different computers and networks. To evaluate some of these difficulties, measurements can be based on protocol-independent benchmarks. Such benchmarks need to be representative, that is, they should reflect the ways in which the protocols are really used. It should also be possible to construct benchmarks at different levels of complexity, ranging from fundamental transactions to simulations of full-scale applications. Protocol benchmarks must also allow the comparison of different protocol implementations. This paper describes the performance measurement results for File Transfer Protocols (FTP), and Remote Procedure Call (RPC) protocol using a benchmark tool, called SPIMS. Comparisons between the performance of ASCII and binary version FTP, and between SUN RPC and Transmission Control Protocol (TCP) RPC are drawn.