Research On Database Connection Pool Technology Based On The Discrete Time Queueing Theory

Database access plays a very important part in the process of programming and developing the modern software systems. The introduction of the database connection pool technology is to solve the reducing system performance issues that a database connection is frequently created and released when the software is running.There is a key part in the database connection pool technology that how to configure the parameter to make the best performance of software systems. The discrete-time queueing theory is introduced to the mathematical modeling and performance analysis of the database connection pool technology. This article studies dependency relationships between configuration parameters and performance measures of database connection pool technology.The research results can provide theoretical bases for the parameter configuration of database connection pool technology.Firstly, we summarize that the database connection pool technology works and derives some system performance measures such as request blocking probability,average number of connections,system utilization,and system throughput etc.The database connections and connection requests of database connection pool technology were abstracted into a server and customer reach based on the thinking of discrete time queueing theory.After that,we complete the preparatory work of mathematical modeling and performance analysis.Secondly, we complete the performance analysis of the loss queueing system model.We model the connection request arrival process as a Bernoulli process and then we assume that the database connection service time is geometric distribution and the number of connections in the initial connection pool is c.The Geom/Geom/c/c discrete-time queueing model with multi-server is built.The theoretical analysis of the model is completed and some system performance measures are given such as request blocking probability,average number of connections,system utilization,and system throughput etc.The numerical examples are intuitively given to discuss the dependence of system performance measures and configuration parameters.Finally, we complete the performance analysis of the mixed system queueing model. On the basis of the Geom/Geom/c/c discrete time queueing mode, a Geom/Geom/c/c+m mixed discrete-time queueing model is built based on the thinking of the connection pool buffer.We calculate the step transition probability matrix of this model and give the recursive formula to solve the steady-state probabilities. Some system performance measures such as request blocking probability, system average waiting time,system utilization,and system throughput etc are given by using the matrix and recursive formula. We intuitively discuss the dependence of system performance measures and configuration parameters through the numerical examples.