Modern network telemetry systems rely on programmable switches to perform the required operations within the data plane in order to scale with the rate of network traffic. These systems create a stream processing pipeline for all telemetry operations and statically map a subset of operations to switch resources. There are two inherent restrictions in this approach: first, the fraction of operations in the data plane decreases with the number and complexity of telemetry tasks. Second, changing telemetry tasks requires rebooting the switch to install a new telemetry task, i.e., no agility.

To address these restrictions, we present Cedar, a first-of-its-kind reConfigurablE Data plAne telemetRy system. Cedar is based on a new Broadcom ASIC with the BroadScan module that offers a wide range of reconfigurable telemetry operations. Leveraging this unique capability, we introduce and explore reconfigurable data plane telemetry systems design issues.

Cedar can periodically re-map different telemetry operations to limited switch resources. This temporal partitioning of telemetry tasks leads to scalability and agility of Cedar. Our experimental evaluations of Cedar demonstrate that both the required switch resources and volume of generated telemetry reports in each period are very small and gracefully scale with the number of tasks.