Optimizing Memory Usage

1. Measuring Memory Consumption

Command	Description
`INFO memory`	used_memory, rss, fragmentation, peak
`MEMORY STATS`	Detailed allocator breakdown
`MEMORY USAGE key [SAMPLES n]`	Per-key estimate
`MEMORY DOCTOR`	Heuristic advice

Directive	Description
`maxmemory 4gb`	Hard cap; triggers eviction policy
`maxmemory-policy`	See section 38

Directive	Effect
`maxmemory-samples 5`	LRU/LFU sample size; higher = more accurate, more CPU
`lazyfree-lazy-eviction yes`	Async free on eviction

Type	Compact Encoding	Threshold Directive
Hash	listpack	`hash-max-listpack-entries` / `hash-max-listpack-value`
List	listpack/quicklist	`list-max-listpack-size`
Set	intset/listpack	`set-max-intset-entries` / `set-max-listpack-entries`
Sorted set	listpack	`zset-max-listpack-entries` / `zset-max-listpack-value`

Tip	Detail
Short names	`u:42:n` instead of `user:42:name`
Hash grouping	Use one hash with many fields instead of many keys

Optimization	Detail
Shared integers	Small ints 0..OBJ_SHARED_INTEGERS reuse one object
Internalize enums	Store enum strings consistently

Command	Description
`HEXPIRE key seconds FIELDS n f1 ...`	TTL on individual hash field 7.4+
`HTTL key FIELDS n f1 ...`	Inspect remaining TTL
`HPERSIST key FIELDS n f1 ...`	Remove field expiration

Warning: Keys > 1 MB cause blocking on DEL, migration, and replication.

Detection	Tool
`redis-cli --bigkeys`	Find largest per type
`redis-cli --memkeys`	Largest by memory

Strategy	Detail
Client-side zstd/lz4	Compress before SET; transparent to Redis
Tradeoff	CPU vs. memory + network

Strategy	Effect
Always set TTL on cache keys	Prevents unbounded growth
`EXPIREAT`	Align to wall-clock boundaries

MEMORY USAGE user:42 SAMPLES 0

Param	Detail
SAMPLES 0	Scan all sub-elements (accurate, slower)

Command	Description
`MEMORY STATS`	Allocator buckets, dataset overhead, AOF buffers
`MEMORY MALLOC-STATS`	jemalloc internals
`MEMORY PURGE`	Reclaim memory back to OS