Real - Time Analysis Of Enterprise Financial Data Based On Memory Computing Technology

It is becoming a bigger and bigger challenge that how to analyze and find valuable information which hides in increasing financial data stored in enterprise computer systems, in real time, when the data size climbs up to a big amount and the flexible data analysis is the requirement. In conventional way, the processing time used to get result from this kind of big data could be a long duration, from tens of minutes to a few days, so that, the requirement of real time could not be met. Some systems, to shorten the long analysis time, generate intermediate aggregation data during transaction or periodically, in which a predefined dimension set must be determined explicitly. In this way, the result of predefined dimension set is retrieved very fast but it is not true when the chosen dimension set is not in the predefined set.In-memory computing is hot technology topic nowadays. It stores data in column oriented way and compresses it. During the run time, the data is loaded into main memory totally. Once it is requested it is passed into CPU cache from main memory directly. Comparing with the way of passing data from disk to main memory, then to CPU cache, this way can reduce the CPU wait time very much. Meanwhile, multi-cores of CPU are utilized to execute the operation in parallel. As result, a single data process time is shortened significantly.Besides the bottleneck of passing data from disk into CPU cache, there is another bottleneck:with the popular enterprise three-tier architecture, the data is always passed between data tier and logic tier. When the data size is big, the performance effect is huge. In-memory computing can manipulate data within data tier, to reduce the performance spoilage because of big data transfer.This paper analysis in-memory computing and its implementation framework, then bring us a real time analytics for enterprise magnanimity financial data with2-tier architecture. At first, the target data to be analyzed is introduced, then the principle of system design is defined, and system function and boundary is defined, and system detail design is defined, after that, the main part of system implementation:model views in data tier, is discussed, finally, testing is applied to prove value of carrying out in-memory computing in real time analysis of enterprise magnanimity financial data.