Telemetry provides invaluable insights into the performance and health of devices and networks. In general, more telemetry data means more insights. However, this comes at the cost of increased bandwidth consumption. Many of the counters inside a device correlate (i.e., the counter values can be recovered knowing the multiplier of the correlation). For example, the number of incoming packets is typically closely correlated with the number of packets sent. Techniques are presented herein that support minimizing the amount of data that is sent by identifying counters that strongly correlate with one another and sending real-time information only for a single ‘main’ counter from such a group and correlation multiplier for the others. A telemetry collector may then restore, offline, all of the absolute values. The benefits from the presented techniques include, for example, bandwidth conservation (as the amount of data that is sent will be significantly reduced), increased security (as a consequence of less data being transmitted, less data will be exposed in case of an attacker capturing the telemetry traffic), simplified issue detection (instances where correlations are not being respected need special attention as something abnormal may be happening with a device, perhaps raising a configured alarm and thus decreasing issue detection times), etc.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Calmîc, Sergiu, "CORRELATION-BASED TELEMETRY COMPRESSION", Technical Disclosure Commons, (July 14, 2021)