Comcast Xfinity Home Oracle Scalability and Replication Architecture

Comcast Corporation is the largest American multinational telecommunications and media conglomerate, the second-largest broadcasting and cable television company in the world by revenue, and the largest pay-TV company, cable TV company and home Internet service provider in the United States. Among its products is Xfinity Home, a home security and smart home automation solution with live video streaming.

The software system underneath Xfinity Home was built and operationally supported by a division of iControl Networks, now part of Comcast. iControl engineered its "Converge" application to store data in the Oracle RDBMS, with distributed active-active replication between Oracle databases provided by Oracle GoldenGate.

RhinoSource worked with the iControl team on the design and testing of Oracle Partitioning for the Converge home security schema, so that exponentially growing IoT data could be stored, retrieved and purged efficiently once expired. We wrote the automation that runs daily to maintain those partitions and drop expired data, along with the online partitioning conversion scripts — which have to run against a live database without disrupting an always-on application.

RhinoSource also designed, configured and tested the Oracle GoldenGate replication supporting the distributed active-active Oracle architecture that the multi-region Converge deployment required.

Then a region of the US was hit with a rash of false alarms and sensor outages. RhinoSource ran statistical analyses on Converge sensor data and correlated the outages with freezing temperatures — leading to the conclusion that a bad batch of sensor batteries had been installed, and that those batteries were the root cause. We also analysed meshed sensor behaviours, which helped the engineering team debug sensor firmware and fix the underlying issues.

Finally, RhinoSource provided a design review for an Apache Cassandra-based Cluster Location Service, allowing the Converge application to scale across multiple active-active data centers.