Generate Latency Percentile Queries for a Payment Processing Service

Problem/Feature Description

A fintech team runs a payment processing service (payment-api) and wants to add P50, P95, and P99 latency panels to their Grafana dashboard. Their Prometheus setup exposes latency as a classic histogram named payment_request_duration_seconds with the standard _bucket, _count, and _sum suffixes. Labels include job, payment_method, and le.

A previous engineer left the following query in a runbook, and the SRE team is unsure whether it is correct:

avg(payment_request_duration_seconds{quantile="0.95"})

The team needs you to generate the correct P50, P95, and P99 latency queries suitable for use in a Grafana dashboard, broken down by payment_method. Write a short explanation next to each query explaining what it computes.

Output Specification

Produce a file called latency_queries.promql containing the P50, P95, and P99 queries, each preceded by a # comment. The queries should be scoped to the payment-api job and broken down by payment_method.