CNR:A Technique for Data Replication Organization in BigData

Internet penetration is increasing every year which ultimately generate the huge amount of data. Data storing techniques along with the underlying networks handling those data storing and processing system are also upgrading. There is always a trade off between the data processing and data transmission due to stacked architecture. Data processing always has a overhead over data transmission in distributed environment. Comparing 2 Tier architecture with 3 Tier architecture provides more data storage capacity. Currently adopted big data storing and processing software solution frameworks (Like Hadoop) are highly dependent upon the networks for its intended working. Data availability and reliability are the main focus for BigData (BD) architecture and frameworks. The performance of Fully distributed deployment uses replication of data for its fault tolerant, availability and reliable working. The number of Frames and packet transferred over the network put a direct impact on the performance of Hadoop with replication. Although the replication of data increases the reliability of data storing but increases the response time and uses extra memory space. Calculating the network overhead due to replication and providing a efficient way to minimize the network load will enhance the performance of BD frameworks. Data compaction can reduce the number the frames transferred over the network. The contribution of the following work is to estimating the number of Frames/Packets need to store block of data and calculating the bandwidth of the network due to stacked based network architecture.