Skip to main content
Insert a large number of sequentially-named key-value pairs and report how long the operation took. Useful for measuring write throughput and observing hash table resize behavior.

​
Syntax

BENCH n

​
Parameters

​
n
integer
required
The number of key-value pairs to insert. Must be greater than 0.

​
Return value

Returns a string describing the result: 
Inserted <n> entries in <elapsed> seconds.

​
Errors

ConditionError message
Wrong number of arguments(error) wrong number of arguments for 'BENCH'
n is zero or negative(error) invalid number

​
Examples

>>> BENCH 100000
Inserted 100000 entries in 0.04 seconds.

>>> COUNT
(integer) 100000
Smaller run to observe behavior: 
>>> BENCH 5
Inserted 5 entries in 0.00 seconds.

>>> GET key0
value0
>>> GET key4
value4

​
Key and value naming

BENCH inserts keys using a fixed pattern:
IndexKeyValue
0key0value0
1key1value1
………
N-1keyN-1valueN-1
All keys are inserted without a TTL (expires_at = 0).

​
Notes

BENCH does not write to the AOF. The inserted keys exist in memory only and will not survive a process restart unless you run SAVE immediately after. This is intentional — AOF writes would dominate the benchmark timing.
Because BENCH inserts many keys rapidly, it is a good way to trigger hash table resizes. Monitor the Resizes and Load Factor fields in INFO output before and after a large BENCH run to observe dynamic resizing.
Run BENCH followed by INFO to get a full picture of hash table health at scale:
>>> BENCH 50000
Inserted 50000 entries in 0.02 seconds.
>>> INFO
...
Buckets : 65536
Load Factor : 0.762939
Resizes : 4
Max chain : 3

Build docs developers (and LLMs) love

Get started for freeTalk to us