P50 Latency
28ms
P95 Latency
45ms
P99 Latency
68ms
Max Latency
142ms
SLA Target
50ms
98.2% met
Pipeline Latency Breakdown
Ingestion
2ms
Feature Fetch
8ms
Feature Compute
5ms
Rule Engine
4ms
ML Inference
11ms
Decision
2ms
Total
32ms
Latency Over Time
Latency Distribution
Component Latency Comparison
Latency by Model
Latency Hot Spots
| Component | Operation | Avg | P99 | % of Total | Trend | Action |
|---|---|---|---|---|---|---|
| ML Inference | XGBoost predict | 11ms | 28ms | 34% | ↑ | |
| Feature Store | Redis lookup | 5ms | 15ms | 16% | → | |
| Feature Compute | Aggregations | 3ms | 8ms | 9% | ↓ | |
| Network Graph | Entity lookup | 4ms | 12ms | 13% | ↑ |