One of the most hyped topics about Big Data is Predictive Maintenance. In production, for example, sensor data from machines are analyzed so that the next scheduled maintenance can be accurately predicted in order to avoid downtime costs. Can this same principle also be transferred to IT infrastructures?
That’s what Big Data Predictive Monitoring does. It collects metrics, at various levels of IT systems (e.g., processor, memory, network, …). The detailed metadata are stored using Hadoop clusters and Timeseries Database. Ultimately, this monitoring serves not only as predictive maintenance, but also to optimize performance and to predict crashes. Unlike from Machines, “self-healing” mechanisms can be also implemented and used.
Francesco Sbaraglia, Big Data Engineer / Manager, Data Reply
Thomas Schmiedel, Data Scientist, Data Reply