Choosing native Aerospike alerts vs Prometheus‑based recording rules to curb latency‑spike noise

Goal: Detect meaningful Aerospike latency or resource issues while minimizing unnecessary notifications.

Constraint: Native threshold alerts fire on instantaneous samples and lack time‑of‑day or maintenance‑window suppression, causing flapping; using Prometheus with recording rules adds exporter overhead but enables sliding‑window averaging and hysteresis for noise reduction.

Questions: Can native alerts be supplemented with external silencing to match Prometheus noise suppression? Is a Prometheus‑based recording rule more reliable for sustained latency events than raw threshold alerts? Which method reduces false positives without increasing operational complexity?