Towards Scalability in Systems with Write Operations in Relational Databases

Abstract

The scalability of systems is essential in scenarios where there is a lot of concurrent users. Some systems have peak access during certain times, others have a lot of concurrent users regularly. Thus, these systems must be able to perform equally as a growing number of requests. The study of system scalability that manipulate database considers read operations in database systems. However, in this paper we present an approach for horizontal scalability on systems that perform write database operations. The proposed approach uses queuing theory and differs from commonly developed solutions like clustering databases and elasticity in the cloud. A case study was conducted with an application of persistent logs, and two versions of the application have been implemented in order to compare the efficiency of the use of queues. Results show the efficiency of the use of queuing theory implemented to allow scalability for applications with a large amount of writing database operations.