An Alternative Approach to In-Memory Data Processing: A case of financial institutions

Abstract

The idea of in-memory based data processing has been present since the 1980s. At that time, the main obstacle was the price of memory, which has been steadily dropping ever since. Today, the technology has evolved on parallel with the big data concept in order to provide a complete solution for modern day business. With the current data growth rate, in-memory based processing systems are more of a necessity then a commodity for a business of any size. Complete commercial solutions exist and among them SAP HANA and Oracle Exalytics are best known. The main reason why more companies are not implementing them is the price. The main goal of this paper is to present the benefits of in-memory data manipulation and provide a solution for businesses which cannot afford the commercial systems. An architecture model is given which is based on the standard three-layer architecture. The business logic layer has been granulated and extended to provide good structure for further development, upgrades and maintenance. The model has been proven as efficient through the implementation of two processes which are common in financial institutions. The results are analyzed on multiple criteria and the most important is time consumption of the execution of each process. Depending on the nature of the process, it is determined that using the in-memory approach the processing speed can be increased even several hundred times. The costs are practically minimal since the solution is aimed at in-house development for critical business processes, compering to the commercial solution of the same type. Therefore, the use of the given model can enable a business to remain competitive without the need to invest an enormous amount of money into this technology.