Mobile operators have long contended with tremendous amounts of data, and in the past relied on summarizing and aggregating data to operate their businesses, so the use and management of large datasets is not new. Telecom data meets the fundamental 3Vs criteria of big data: velocity, variety, and volume, and should be supported with a big data infrastructure (processing, storage, and analytics) for both real-time and offline analysis.
Telecom big data spending includes distributed storage and computing Hadoop (and Spark) clusters, HDFS file systems, SQL and NoSQL software database frameworks, and other operational software. Telecom analytics software, such as for revenue assurance, business intelligence, strategic marketing, and network performance, are considered separately. The evolution from non-machine learning based descriptive analytics to machine learning driven predictive analytics is also considered.
Revenue, or operator spending, is broken down by big data infrastructure (hardware, software, and services) and telecom analytics (machine learning and non-machine learning, and descriptive and predictive). Forecasts are through 2021. This research deals primarily with mobile telecom operators, but some operators include wireline operations, which are serviced by the same big data and machine learning infrastructures.
A selection of vendors in the telecom space are also surveyed.