Workload prediction through Fast Data Streaming

In modern IoT use cases, data communication is getting more and more complex. IoT scenarios are based on an increasing amount of edge devices, which use a large variety of data types and communication protocols. These dynamic and data-intensive edge processing scenarios challenge communication infrastructures, which need to become more flexible and adaptive to new situations.

Simple communication brokers “only” enable the transfer of data, without understanding the traffic semantically or syntactically. Therefore, methods for intelligent processing (ensuring the understanding of real-time situations) are necessary. In PrEstoCloud, this will be realized through a workload predictor. This workload predictor enables a better understanding of the data transferred over a broker. It predicts the workload in the near future and enhances the concept of the broker from “just brokering” into managing how the brokering can be done. The workload predictor logs all data sent over the broker, enabling new services like:

  • Controlling the work of a broker
    • discovering the need for a new broker (distribution)
  • Monitoring the usage of a broker
    • detect active periods and the overload of the broker

A previous solution for the workload predictor has been developed by the PrEstoCloud partner NISSATECH. This solution is based on an open source software. As the usage of open source software often goes along with missing support opportunities for customers, Software AG aims to present an alternative realization of the workload predictor by using Software AG´s products Mashzone NextGen and APAMA – see the figure below. APAMA is a platform for performing streaming analytics in real-time. Mashzone is a visual analytics software which enables to explore and analyze streaming data fast by visualizing them on Dashobards. Using APAMA and Mashzone together within PrEstoCloud’s architecture allows to instantly get insights from big fast data streams from IoT devices and to predict what workload is likely to happen next.

The new variant will soon be presented in a demonstration.