Design And Implementation Of Computer Availability Modeling And Assessment Tool

High-end fault-tolerant computer systems are widely used in key areas such as finance, telecommunications, etc. They are different from the ordinary transaction processing systems. High-end fault-tolerant computer systems have not only strong system process capacity but also high system availability. Otherwise the systems failure may cause devastating disaster. Therefore, how to evaluate the availability of high-end fault-tolerant computer is very important. The paper develops a computer availability modeling and assessment tool. This tool supports the assessment and measurement for high-end fault-tolerant computer through the evaluation of the target system availability, performability and other indicators.This paper firstly gives a survey on the status and the existing tools of evaluating the availability both at home and abroad. And then the evaluation indicators and their testing process are come up with. They meet the need of availability measurement for high-end fault-tolerant computer. They overcome the separation of fault-tolerant indicator and system performance indicator, the singler evaluation index and so on in traditional evaluation system. On this basis a computer modeling and assessment tool is developed to test and analysis the system availability, system performability, the threshold and measured value of MTBF(Mean Time Between Failure) of the parts. Based on the above demand for different types of testing we divide the tool into the following seven functional blocks.The master control module is the console of the tool. It is responsible for scheduling the implementation of other functional modules, provides database access interface and GUI. Drawing system similar RBD(Reliability Block Diagram) module draw subsystem block diagram, describe the system redundancy strategy and draw automatically the system similar RBD in the user interface one by one. System availability analysis module analyzes various subsystems which constitute the target system and system-wide availability based on combination model and Petri nets. The sensitivity analysis of component failure module analyzes the influence on system availability with the failure rate changes of specified component. The corresponding analysis graph and analysis chart are given. The sensitivity analysis of workload strength module analyzes the influence of different levels of workload intensity on system availability. Then it draws the sensitivity curve of workload intensity. The value of component MTBF analysis module consists of two sub-modules. They analyze separately the threshold and measured value of MTBF of the parts. System performability analysis module analyzes the steady-state probility and the average processing performance when target system is in normal, degraded and failure, three state modes. Based on this we can analyze system performability and draw the target system performance-time curve.Finally, this paper measures and evaluates system availability and other related indicators of HP Superdome and HEFT in the research, two high-end fault-tolerant computers. Meanwhile the result has been detailed compared and analyzed to provide some suggestions for the improvement of HEFT system prototype. These experiments also indicate this tool has basically met the availability evaluation demand for high-end fault–tolerant computer.