Why is Redis an in-memory DB?

Redis is called an in-memory database, because all data is stored and processed right in RAM, rather than on disk like classic DBMSs.

#The core idea

Redis works entirely out of RAM, to provide:

• instant access to data (access time in microseconds),
• high throughput (millions of operations per second),
• low latency even with a large number of clients.

Disk is used only for backups or persistence, not for the primary work with data.

#How this is set up

1. Every data structure (String, List, Set, Hash, ZSet, etc.) lives in RAM.
2. Every operation (GET, SET, INCR, etc.) runs directly on the objects in memory.
3. So data isn't lost on restart, Redis can:

• periodically save a snapshot (an RDB snapshot) to disk,
• or log every write (AOF, Append Only File). On restart, Redis loads that data back into memory.

#Why this matters

• RAM is tens of thousands of times faster than disk.
• There's no slow I/O, reads and writes go directly against RAM.
• That makes Redis ideal for caches, queues, sessions, tokens, counters, and real-time analytics.

#But Redis isn't a "pure cache" for all that

Although Redis runs in memory, it can be made persistent:

• With RDB or AOF enabled, data is saved and restored after a restart.
• Redis can also run in replication and cluster mode, which makes it resilient to failures.

#Advantages of the in-memory approach

• Speed: millions of operations per second.
• Minimal latency (<1 ms).
• The ability to perform complex operations right in memory (aggregations, sorting, queues).

#Limitations

• The data volume is limited by how much RAM is available.
• Once it fills up, Redis applies an eviction policy, removing old keys (e.g. volatile-lru, allkeys-lru, and others).
• Data is lost on a power failure, unless persistence has been configured.

So, Redis is an in-memory key-value store, where disk plays a supporting role, and all the work with data happens in RAM to deliver maximum speed.