By Nitin Kumar, Amir Leventer & Asaf Matatyaou, Harmonic, Inc
Access systems such as Cable Modem Termination Systems (CMTS) have traditionally been monitored using tools that leverage CLI and SNMP interfaces. These same interfaces are also used to gather live information while troubleshooting issues in the field. The data exposed by these interfaces are limited by the standard set of commands and MIBs supported by the system. With the move to a distributed access architecture (DAA), deployments can support much higher scale, and their software-centric designs make available richer data useful for vendors in operating, debugging, and optimizing the systems, but access to this increased amount of data is bottlenecked by the limited performance and limited extensibility of CLI and SNMP interfaces. Newer system designs support streaming logs and telemetry to overcome these limitations, and this paper looks at how these features not only overcome the limitations of traditional interfaces but also enable more efficient monitoring and troubleshooting workflows. We will show the building blocks of a system that implements streaming logs and telemetry. The performance and security improvements offered by such a system will be noted. We will illustrate some monitoring features and describe something we call “time-travel debugging” — using the streamed data for easier troubleshooting.
Finally, we will give an overview of using the data for advanced analytics and machine learning, specifically for faster root cause analysis of field issues.