The computing and communication power in the cyber-physical world is expanding greatly. As a result, a lot of data is generated to manage these activities. Big data has four primary challenges: volume, variety, velocity, and authenticity. Some storage-based data processing systems, like Hadoop, manage volume and variety. However, the speed and accuracy of processing such a vast volume of data require an overly complicated process. In this paper, we'll put into practice a system that can deal with huge volumes, varied patterns, and the speed of data. To extract valuable information from the data stream, we'll use correlation analytics and data mining. The system must be able to process data in real time, using an event processing engine like Esper that can generate various events using different language queries. Storm, which uses topology, is used to capture real-time data and for straightforward filtering of that data stream. Apriori and FP-Growth are two separate algorithms that are used for correlation and mining. Data centers all across the world are now using Apache Hadoop. The common programmer can now use parallel processing. It is essential to convert current data mining methods to the Hadoop platform as more data centers support it to maximize the effectiveness of parallel processing. The tendency of moving current data mining algorithms to the Hadoop platform has grown widespread with the advent of big data analytics. We examine the present migration activities and problems in this survey research. The reader's suggestions for solutions to the present migration difficulties will be guided by this essay.