OpenTelemetry, an open-source project, enables the instrumentation, generation, collection, and exportation of observability data. ThousandEyes as part of Cisco collects 30000 data per second. Currently, ThousandEyes network test data is only accessible via Rest API. Nowadays everyone is talking about Opentelemetry and real time observability. To simplify the integration with ThousandEyes, we have developed a system based on OpenTelemetry to export network test data. It allows customers to correlate ThousandEyes data with their data.
Nevertheless, with this wave of adoption also came a wave of concerns and risks that simultaneously prevent many companies from onboarding the technology. Privacy, security, reliance on third party and costs, are all reasons that have slowed down the adoption of AI for internal processes at a corporate level.
Our OpenTelemetry architecture is mostly built using the existing OpenTelemetry Collector components. The system is capable of processing and exporting 3k data points per second. Nevertheless, we had to develop some extensions, which we are planning to contribute to the community.
In this session, we will present the initiative for adopting OpenTelemetry, including a walk-through of the architecture, some of the custom components we have developed, and a demo at the end about how easily customers can export the network test data to any observability backend.