Backpressure & Flow Control
The fabric manages telemetry flow so a burst never drops safety-critical data, prioritizing channels and applying backpressure to the rest.
When data outruns the pipe
Sub-microsecond sampling across 60+ ports can, in a burst, generate data faster than a link or store can absorb. The fabric handles this deterministically: it prioritizes safety-critical and control-critical channels, buffers within bounds, and applies backpressure to lower-priority flows rather than dropping data at random.
Priority tiers
- Tier 0: safety and control features that feed L1 and the failsafe context - never dropped.
- Tier 1: diagnostics the twin's predictive shadow needs live.
- Tier 2: full-fidelity raw series for the archive, buffered and shipped as bandwidth allows.
- Tier 3: bulk documents and non-time-critical records.
Backpressure, not loss
When a downstream stage saturates, the fabric signals upstream to slow or buffer lower tiers, keeping the critical path clear. Because the archive tolerates a bounded delay, Tier 2 data can queue at the edge and drain later without loss, while Tier 0 flows unobstructed. What cannot be buffered within bounds is recorded as a gap in lineage, never silently lost.
Determinism matters
In a control system, unpredictable data loss is a hazard. Flow control is designed so that under any load the critical telemetry keeps its timing guarantees, which is what lets the sub-10 microsecond boundary and the twin's shadow stay valid during exactly the disruptive events when data rates spike. It serves both machines and is a design specification for machines not yet built.