This article explains how Redis RDB snapshots work, compares the blocking SAVE command with the asynchronous BGSAVE command, outlines their impact on performance, and provides best‑practice guidance on snapshot frequency and trade‑offs.
Redis achieves high speed by keeping data in memory, but to avoid data loss it offers two persistence methods—RDB and AOF; this article explains RDB’s snapshot mechanism, the SAVE vs BGSAVE commands, copy‑on‑write handling, automatic save configuration, and the internal structures that drive periodic backups.
The article analyzes a recurring one‑second pause in a Redis 4.0 master‑slave setup, identifies the periodic BGSAVE fork operation as the root cause through metrics like latest_fork_usec, and presents mitigation strategies such as memory limits, activedefrag, and migration to a lower‑RSS instance.
