The Petabyte Telemetry Archive
A petabyte-scale archive retains raw and reduced telemetry at high fidelity, so any shot on either machine can be re-examined byte-for-byte.
Retain everything that matters
Sub-microsecond sampling across 60+ ports generates telemetry at a rate that accumulates to petabytes. The archive retains it at high fidelity — raw counts where forensics demand it, reduced features everywhere — so no shot is lost to summarization. A disruption in the breeder or a plug collapse in the burner can be reconstructed exactly, years later.
Tiered retention
- Hot tier: recent shots at full raw fidelity for fast forensic access.
- Warm tier: reduced features and validated channels for training and analysis.
- Cold tier: compressed long-term retention for lifetime and regulatory records.
- Raw is never overwritten by corrections; calibrations apply at read time.
Why raw fidelity is non-negotiable
Corrections and calibrations improve over time. Keeping raw counts means an archived shot can be re-reduced with a better calibration or a new feature definition — the training data for the twin gets better without new experiments. Lossless compression shrinks storage without discarding information; lossy reduction is applied only where information theory shows nothing is lost.
Indexed for retrieval
The archive is indexed by shot, machine, time, and physics regime, and cross-linked to the shot database and the vector database so a human or a copilot can find the right shot by description, not just by ID. Every record carries full lineage. The archive serves both machines under one schema, versioned in the registry.