An evaluation and performance comparison of different approaches for data stream processing

Abstract: In recent years the demand of faster data processing and real-time analysis and reporting has grown substantially. Social networks, internet of things, trading are among others, use cases where data stream processing has a vital importance. This has led to the emergence of several distributed computing frameworks that can be successfully exploited for data stream processing purposes. This project aims to examine a number of them, their architecture and key features. First, all the open source frameworks were found and studied. Based on the approach they follow, two of them were selected to be further analyzed and presented in detail. In the final part of the project a telemetry data monitoring application was applied using both frameworks on a computing cluster. The aim of that experiment was to illustrate how those two different approaches would perform in terms of exploiting the clusters resources as they scale out.