Scalability — FUTURE

Possible goal: grow to multiple users/connections without compromising streams and jobs. The current system makes no such claim.

Before changing the architecture, measure:

• concurrent connections and requests/s;
• stream and Range concurrency;
• TTFB, p50, p95, and p99 by route;
• memory/CPU by format and size;
• PostgreSQL capacity, connections, and latency;
• bandwidth and cache hit ratio;
• queue depth/latency/failure;
• sync and enrichment time.

Possible phases: reproducible baseline → remove a proven bottleneck → separate shared state → isolate workers → test multiple instances. Do not choose PostgreSQL/Redis/R2 based on expectation alone.