As Redis aims to serve its complete dataset in memory, having a sufficient amount of RAM is paramount to achieving high levels of performance. The Redis FAQ offers some estimates on how the RAM footprint is affected by differing levels of data length and width:[5]
- Empty instance — about 1 MB of RAM
- 1 million small key/value (string) pairs — about 100 MB of RAM
- 1 million key/hash pairs (object with five fields) — about 200 MB of RAM
These are estimates and should only be used as guidelines. Always thoroughly test and estimate with a sample of your own data to accurately predict the RAM footprint prior to deploying in a production environment.
