Abstract
Latency depends on your setup — particularly the complexity of your rule framework and whether you're on a Standard or Premium SLA.
Pingwire offers the following latency performance, measured as the 99th percentile (p99) on a monthly basis:
| Service Level | Base System Latency Limit(p99) | Rule Latency Limit (p99) |
|---|---|---|
| Standard | 750 ms | 750 ms |
| Premium | 500 ms | 500 ms |
The total latency for a request is calculated by adding:
- The base system latency
- The latency of the rule branch involved in the decision
If a rule is nested within another rule, their latencies are added together. If multiple rule branches are evaluated in parallel, only the longest branch is counted.
Example of Latency Limit Calculation (Premium):
- Rule 1: 300ms
- Rule 2 (nested under rule 1): 300ms
- Rule 3 (nested under rule 2) with no recommendation or signal involved: 300ms (not counted)
- Rule 4: 300ms
Calculation: First branch total latency = 600ms, second branch = 300ms. Overall rule tree latency = the maximum of branch latencies, i.e 600ms + Base System Latency 300ms = 900ms