A Real time Streaming Video Analytics Stack for the Edge

Edge computing is a process which brings complex computation closer to the location where it is needed, to improve response times and save bandwidth. The number of use cases for edge computing is growing exponentially and includes everything from enabling smart cities to improving response time for emergency medical situations. Currently many edge/IoT implementations enable vendor lock in and work only with a select set of  applications and devices. To make building cloud native edge device systems in a reliable and scalable way a flexible middleware stack using the power of open source needs to be available. 

The real time streaming video stack demo proposes a completely open source solution for an end to end messaging and persistent storage stack. For the core messaging platform It uses Apache Kafka, deployed on openshift via the Strimzi operator, to provide low latency and high availability with the streaming video. On the application side, the serverless framework knative efficiently digests and serves the data back to the user. Real time analytics is done via a Tensorflow serving instance, while  data persistence is handled using Ceph object storage, providing a modular, user configurable, and scalable storage entity. This demo is part of a larger project which involves making an Eclipse Che and VScode plugin which will make deploying the above tools and creating edge systems much easier for the developer

Red Hat Interns: Andrew Stoycos, Aditya Pradip Kadem