Study of Optimized Window Aggregate Function for Big Data Analytics

Abstract

Background: A Window Aggregate function belongs to a class of functions, which have emerged as a very important tool for Big Data Analytics. They lend support in analysis and decisionmaking applications. A window aggregate function aggregates and returns the result by applying the function over a limited number of tuples corresponding to current tuple and hence lending support for big data analytics. We have gone through different patents related to window aggregate functions and its optimization. The cost associated with Big data analytics, especially the processing of window functions is one of the major limiting factors. However, now a number of optimizing techniques have evolved for both single as well as multiple window aggregate functions. Methods: In this paper, the authors have discussed various optimization techniques and summarized the latest techniques that have been developed over a period through intensive research in this area. The paper tried to compare various techniques based on certain parameters like the degree of parallelism, multiple window function support, execution time etc. Results: After analyzing all these techniques, segment tree data structure seems better technique as it outperforms other techniques on different grounds like efficiency, memory overhead, execution speed and degree of parallelism. Conclusion: In order to optimize the window aggregate function, segment tree data structure technique is a better technique, which can certainly improve the processing of window aggregate function specifically in big data analytics.