Society is becoming increasingly more instrumented and as a result, organizations are producing and storing vast amounts of data. Managing and gaining insights from the produced data is a challenge and key to competitive advantage. Analytics solutions that mine structured and unstructured data are important as they can help organizations gain insights not only from their privately acquired data, but also from large amounts of data publicly available on the Web. The ability to cross-relate private information on consumer preferences and products with information from tweets, blogs, product evaluations, and data from social networks opens a wide range of possibilities for organizations to understand the needs of their customers, predict their wants and demands, and optimize the use of resources. This paradigm is being popularly termed as Big Data. In the existing system, the system can dynamically balance load via tasks redistribution/stealing accord-ing to the profiled load from the previous iterations. However, they cannot support the computation de-composition of straggling FEPs. In this project the system proposed a general straggler-aware execution approach, SAE, to support the analysis service in the cloud. It not only parallelizes the major part of straggling FEPs to accelerate the convergence of feature calculation, but also effectively uses the idle time of computers when available. In order to reduce communication cost, SAE also aggregates these messages that are sent to the same node. At the first iteration, SAE only divides all data objects into equally-sized partitions. Then it can get the load of each FEP from the finished iteration. SAE can get a speedup up to1:77 times against PowerGraph on Adsorption algorithm.