In-Memory Database vs. On-Disk Database

Where the data physically lives changes almost everything about a database's behaviour.

An in-memory database stores its data primarily in the main memory (RAM) of the server, which gives very fast, low-latency access. An on-disk database stores data on disk, which is slower to read but survives a power cut.

In-Memory

Removing disk access from the read path is what makes these databases fast. There is no seek, no page fetch, nothing to wait for. Response times drop to a level a disk-based system cannot reach.

Redis, Memcached and Amazon Elasticache are the usual examples.