Staleness Data Retrieval from Multi-Tiered and Replicated Clusters

Abstract

The tsunami of data spans over the network is now coined with the term “Big Data”. The problems faced by Big Data during data acquisition are Privacy, Security, Complexity, Heterogeneity, Storage etc. Which data to keep and which data to discard is also a big question regarding Big Data since some data are not stored in structured format. In Social networking like Facebook, twitter, Whatsapp etc, the data communication takes place only in unstructured format and so later the conversion and interpretation of those data into presentation of results is a very clear bottleneck [1]. So Big Data analysis now pulls every aspect of our usage like retail marketing, mobile services, financial services, life sciences, production and sales. So the real challenge is to get a maximum out of the data available already and predicts the nature of data to be collected as per future need. Mostly read transactions outlay write transactions over the storage of data. So this paper proposes a new methodology “Read Polling Algorithm” to perform the read operation effectively and efficiently for retrieving correct data to the intended user..