Study Of Data-Access Optimizations In The Financial Field

With the development of information technology, the model of data access in thefield of finance has changed enormously. Concretely, the requests are submitted by thedata-access middleware rather than with sending the SQL statements. Without sendingdata to database, data-access middleware can reduce the burden of back-end databasesignificantly. However, the throughput performance of real-time traffic ofdatabase-access middleware is easy to become the bottleneck of real time. So, it directsthe future studies in the area of computer applications that how to provide highefficient data access, and I also choose it as the focus of my research.The existing data-access middleware have a low real time capability, whichencourage me to carry out research on data-access middleware, such as thedevelopment, architecture and key technology of middleware. This thesis focus on twokey technologies: resource pooling management and real-time task scheduling.Resource pooling management is a resource management method, adopting thetechniques of process pool and database connection pool, which can avoid creatingprocesses and connections frequently by pre-creating a certain amount of processesand connections. Based on the research on resource pooling management and existingalgorithms, a new algorithm based on historical records of resource poolingmanagement is proposed.The goal of real-time task scheduling is to ensure that each task can be satisfactoryfor its time constraints and respond to external requests promptly. A taskclustering-based minimum laxity first algorithm is put forward based on existingreal-time task scheduling algorithms and priority scheduling algorithms.Because the real-time task scheduling plays an important role in database-accessmiddleware，an special simulation program is designed and a conclusion that the newalgorithm has better performance is made.