3
AC
Diagnostics/Load

Load, Scale & Capacity

Production capacity analytics and controlled load testing with hard safeguards. Never unauthorized third-party testing.

Peak concurrent

312

Queue depth

14

Worker util.

71%

Saturation warnings

3

Concurrent calls

247

Calls per second

4.2

Sessions per region (us-east)

98

Queue depth

14

Worker utilization

71%

ASR concurrency

180/250

LLM rate-limit headroom

22%

TTS concurrency

140/200

API rate limits (tools)

65%

Telephony capacity

400 ch

SIP-channel capacity

380/500

Tool throughput

90 rps

Database load

58%

Error rate under load

1.2%

Latency under load (P95)

2.1s

Cost under load ($/min)

$1.84

Concurrent sessions & queue (24h)

Forecasts & recommendations

Capacity forecast

At current growth, SIP-channel capacity reaches 90% within ~18 days of peak-season traffic.

Saturation warning

LLM rate-limit headroom at 22% — expect 429s if outbound survey volume doubles.

Resource recommendations

+50 ASR concurrency · raise tool API quota · pre-warm TTS pool for peak 10–11am.

Expected peak analysis

Peak concurrent forecast tomorrow: 380 (95% CI 340–410).