Redis is an in-memory key-value database introduced in 2009 by Salvatore Sanfilippo to solve scalability issues at his startup LLOOGG. It delivers low-latency operations by holding all data in memory and using the fork system call to enable concurrent client service and disk persistence. It supports two persistence methods—RDB snapshotting and AOF journaling—with journaling introduced in version 1.1 and considered safer. Default configuration limits data loss to a few seconds upon full system failure. Redis supports master-replica replication with hierarchical trees and introduced clustering in April 2015 with version 3.0. It grew into one of the world’s most popular databases for caching, message brokering, and primary storage. In December 2024, Sanfilippo returned to Redis and implemented the Vector Set data structure for vector similarity search.
workloads where even two-second data loss is unacceptable
teams without capacity to tune replication or clustering
Who should study it, and who should skipHide who it is for
The written brief2 min read
What it is and the problem it solves
Redis is an in-memory key-value database prototyped in 2009 by Salvatore Sanfilippo to solve scalability bottlenecks in his startup LLOOGG’s real-time web log analyzer. It addresses slow disk-bound databases by keeping everything in RAM and serving low-latency operations.
How it works
Redis holds all data in memory and uses the fork system call to duplicate the process: the parent serves clients while the child persists data to disk. It offers two persistence methods: RDB snapshotting and AOF journaling. Journaling was added in version 1.1 and is considered safer. By default, Redis writes to disk every two seconds, limiting data loss to a few seconds on full system failure. Replication supports hierarchical master–replica trees. Clustering arrived in April 2015 with version 3.0.
What works
Low-latency reads and writes work reliably due to in-memory operation and efficient single-threaded event loop. Master–replica replication works with hierarchical trees. AOF journaling provides safer persistence than RDB snapshots. Redis is widely adopted for distributed caching and message brokering because these patterns align with its strengths.
What does not
Redis does not guarantee durability by default. Its fork-based persistence creates memory overhead and can stall under large datasets. Journaling avoids some risks but adds latency and disk I/O pressure. Clustering is complex to operate and was not part of the original 2009 design. The Vector Set—added in December 2024—is new and unproven at scale.
What it changes
Redis changed how developers think about database roles: it blurred the line between cache and primary store. It made in-memory data structures viable for production messaging and real-time analytics. It shifted persistence expectations from ACID guarantees to ‘a few seconds of loss’ as acceptable for many use cases.
Is it worth your time
Yes—if you need low-latency reads and writes for caching, message brokering, or primary storage, and can accept the trade-offs of in-memory operation and persistence gaps. It demands careful configuration to avoid data loss; default settings are not production-safe for critical workloads.
