What is data persistence & why does it matter?

Persistence in data storage refers to the continuance of an effect after its cause has been removed, meaning that data survives even after the process that created it has ended. This is achieved by writing data to non-volatile storage. There are four main design approaches for data stores: in-memory, update-in-place, commitlog-based, and snapshot-based systems. In-memory systems provide high speed but limited data sets and lack persistence. Snapshot-based systems offer periodic snapshots to disk but can lose updates within the interval. Commitlog-based systems store data immediately and provide durability with every write. Cassandra implements a commit-log based persistence design, allowing for tunable levels of durability to balance safety and performance. This makes it a logical choice for systems requiring durable, performant data storage.