Performance analysis and evaluation of complex EIS based on structure graph

Past research on high performance for enterprise applications has concentrated on CPU speed, memory capacity, networks bandwidth and exploitation of parallelism, but has, until very recently, neglected the interior structure of various applications and services in complex Enterprise Information Systems. As the enterprise IT system becoming more complex, more seriously its interior structure effects on its performance. However, there are few appropriate approaches for high performance of EIS based on the structure analysis. This paper presents a new method based on system structure graph, and then gives the essential elements of the structure graph-application nodes and link arcs, based on the discussion of the relationship between system structure and performance. The analysis approaches and evaluation index system are described with emphasis. For performance analysis and evaluation of complex IT systems, the authors propose several points--load, density, bottleneck and layer, and two aspects of indexes-structure rationality and resource allocation, which could give more instruction for fault diagnosis and optimization. Finally, the. merits and demerits of the evaluation system, and the future development are discussed.