Latency Troubleshooting Interview Questions
How would you troubleshoot high Redis latency? — spoken sample answer for Indian interviews.
- 5Questions with answers
- 3Difficulty levels
Questions (5)
Browse beginner, intermediate, and advanced questions with answers — hide them when you want to self-test.
How would you troubleshoot high Redis latency?
I would not start with 'restart Redis'. I would check slowlog for KEYS, big SMEMBERS, or huge SORT. I would look at used_memory and whether we are evicting hard. I would hunt big keys. I would check client pools and network. RDB or AOF rewrite forks can pause a single-threaded command path. One hot key can melt one shard. Fixes I have used: SCAN not KEYS, split big keys, pipeline, move heavy aggregations off Redis. I would watch commandstats, ops per second, and blocked_clients. Redis commands are mostly one thread. More CPU cores do not magically make KEYS * fast.
A command is slow. First look?
SLOWLOG, big keys, KEYS/LRANGE, fork, CPU steal. I would not restart first. Latency is usually us, not Redis is slow. Evidence first.
Why KEYS * takes down prod?
Full keyspace scan on the main thread. Use SCAN. Block KEYS in prod ACLs. MONITOR is also dangerous. I have seen this incident.
Client latency versus Redis latency?
Compare SLOWLOG with app metrics. If Redis is fine, look at pool, DNS, too many round trips. Pipeline. Two graphs. Two clocks.
Pool too small versus Redis overloaded?
Small pool: threads wait, Redis CPU idle. Overloaded Redis: CPU high, SLOWLOG full. Doubling the pool when Redis is at 90 percent CPU makes it worse.
Practice with AI mock interviews
Run Redis mock interviews with AI follow-ups, instant feedback, and analytics on AiLx.
Free to start · No credit card required