Modern Big Data Architectures

CHAPTER 8Physically Distributed Systems – Mobile Cloud, Internet of Things, Edge Computing

In Chapter 5 we have analyzed architectures for performing large scale computations in the cloud, primarily in the Infrastructure as a Service (IaaS) model. This works well for systems that can be broken down into a number of services and accessed by light-weight, typically web-based, clients. However, things get more complicated when parts of our system are physically distributed devices, which can be equipped in unique sensors and actuators, but have limited computational capabilities. In the age of big data we want to take advantage of the data generated and stored across our infrastructure and be able to take informed decisions at any physical point of our networked resources.

In other words, this chapter takes a closer look at how modern big data architectures distribute computing in order to optimize efficiency, latency, and other KPIs. On the one hand, we look deeper into the cloud technologies, allowing for offloading and scaling of large computations. On the other hand, modern ubiquitous environments and the growing power of distributed devices enables computation to move towards the network edge, which minimizes the need for data transfers as well as reduces the latency. By utilizing both we can build powerful solutions for big data applications.

The first example of such setups are mobile cloud systems, which enable users of mobile devices (typically smart phones) to use rich ...