Systems Reasoning
Compact Contracts: Compressing Telemetry for Cross-Layer Reasoning
A compression layer that preserves cross-layer diagnostic context while reducing raw MELT telemetry volume by roughly 6,000×.
Mohan Murari, Raghav Balasubramaniam — Raincurve Research
Abstract
Continuous reasoning across a large operational environment means reasoning over massive volumes of metrics, events, logs, and traces. Compact Contracts convert raw MELT data into structured representations that preserve the cross-layer information needed for diagnosis while reducing token volume by approximately 6,000× and lowering average inference cost by roughly 47%.
Compression method
The target is not generic smaller text. Each contract preserves relationships between network, compute, cloud, application, and infrastructure signals specifically for the downstream reasoning pipeline, discarding redundant volume that does not contribute to correlation or root-cause tracing.
Cost impact
A reduction of this scale changes the economics of reliability intelligence: continuous, system-wide reasoning can operate 24/7 rather than only on demand, on a sampled subset, or once an incident is already visible.
Conclusion
Compact Contracts are the representation layer that makes Raincurve’s broader cross-layer reasoning platform economically viable without sacrificing diagnostic context.