IntDB: Towards a Spatiotemporal Database for In-band Network Telemetry

In-band Network Telemetry (INT) provides finegrained, path-level network visibility by embedding measurement data into live packets. However, the nested, spatiotemporal nature of INT data poses significant challenges for traditional telemetry storage systems, which are primarily optimized for time series data. In this paper, we present IntDB, a spatiotemporal database for INT. IntDB introduces a path-oriented data model that preserves end-to-end flow structure and supports efficient multihop correlation and querying. Experimental results show that IntDB outperforms existing solution in both writing scalability and query performance, making it a practical foundation for INT-based network measurement.