Verge DB: A Database for IoT Analytics on Edge Devices

Abstract

New infrastructure is needed to gather, store, and analyze massive amounts of time-series data generated by the explosion of Internet of Things (IoT) applications. We anticipate that considerable data processing will need to occur at the edge of the network in order to achieve these scalability requirements. In this work, we introduce VergeDB, a database designed for adaptive and task-aware compression of Internet of Things data, which treats complicated analytical tasks and machine learning as first-class operations. Depending on the context, VergeDB may act as either a lightweight storage engine that compresses the data depending on downstream duties or as an edge-based database that handles compression as well as in-situ analytics on raw and compressed data. VergeDB will make choices to improve speed, data compression, and downstream job correctness by making the most of the available CPU resources, storage space, and network bandwidth.