Mobile Log Aggregation: Token-Bucket Rate Limiting & Priority Sampling

Mobile apps generate far more log volume than a metered network or a collector's storage budget can absorb, so remote logging pipelines admit events through a token-bucket rate limiter rather than shipping everything. This simulator renders eight mobile devices streaming DEBUG/INFO/WARN/ERROR/FATAL events toward a central collector in 3D: a token bucket refills continuously at a bandwidth budget you control, and every event that arrives when the bucket is empty is dropped. Switch between priority-aware admission — which always lets crash-level FATAL and ERROR events through first — and naive uniform sampling, then trigger a crash spike to see the two strategies diverge on the one metric that actually matters: how many crash reports survive under load.