CENTILE: A Telemetry Foundation Model Evaluated by the Decisions It Drives
2026-08-03 • Networking and Internet Architecture
Networking and Internet ArchitectureMachine Learning
AI summaryⓘ
The authors created a model called \sys that learns from an operator's event data to predict future network and system behavior without needing separate tools for each task. Unlike past methods, \sys uses event-driven data with irregular timing and can make predictions on different future periods all at once. Tests showed \sys helps improve decision-making for high-performance computing and network management more than existing forecasting methods. It works well even when used without extra training on new data or different areas.
telemetrygenerative modelevent-driven streamsconditional quantileshigh-performance computing (HPC)network provisioningforecast horizonbackfillingzero-shot transferpretrained model
Authors
Zifan Zhang, Zhichao Hou, Tingxiang Ji, Yuchen Liu
Abstract
Modern computing and networking infrastructure emits telemetry continuously, yet operators convert it into decisions with a separate predictor per task, entity, and horizon. One generative model, pretrained once over an operator's own event streams, could replace this fleet, an approach that already scales to high-cardinality streams in recommendation systems. However, point-forecast error on operational telemetry saturates near simple last-value baselines, so lower error alone need not improve the decisions it feeds. To close this gap, we present \sys, a generative foundation model for network and systems telemetry, evaluated by replaying the decisions its calibrated conditional quantiles drive. \sys treats heterogeneous telemetry as event-driven, irregularly timed entity streams and serves flexible forecast horizons in a single pass, requiring no future timestamps. To our knowledge, \sys is the first pretrained telemetry model to improve both HPC scheduling and network provisioning decisions under replay, its runtime estimator transferring zero-shot across months and its pretrained weights across domains from hours of target data. Extensive experiments on HPC job logs and network traffic confirm that \sys lowers the mean bounded slowdown of backfilling by up to approximately $77\%$ over deployed user estimates and roughly halves the deployed rule's violation rate. Our code is available at https://github.com/ZzZTripleZzZ/all-in-one.