Network Infrastructure for High-Performance Big Data Analytics

Abstract

Big data is becoming a major focus area for present-day organizations. But the data types which form a part of big data need drastic changes in all aspects of IT hardware in order to store, process, and transmit them. Several IT infrastructure transformation steps are taken in an attempt to handle big data efficiently. However, whenever it comes to networks, transformations in the network architecture are very limited due to the fact that it is an interdependent family of components and hence changing one component will require modifications to all other components of the network as well. The traditional network infrastructure has several shortcomings which make them unsuitable for transmission of big data.

It is the need of the day to devise techniques to make present-day networks suitable for handling big data. Some of the techniques which could be used are network virtualization, software-defined networks (SDN), two-tier leaf spine architecture, and network functions virtualization. Each of these techniques is explained in detail in this chapter.

Most of the conventional networks are not optimized for transfer of big data. Protocols like TCP which are used for long-distance transmission of big data have a lot of drawbacks. A novel approach for transfer of big data efficiently using TCP/IP protocol using a technology called FASP is also discussed in this lesson. Some of the implementation aspects of FASP are also discussed in this lesson.