Write-through caching waits for the database before acknowledging a write. That keeps the write path simple, but every write pays database latency.

Write-behind caching, also called write-back caching, changes the contract. The cache accepts the write into a fast memory-backed buffer, acknowledges the application immediately, and persists the write to the database later from a background worker.

This can reduce write latency and smooth traffic spikes. It also opens a window where the application has been told the write succeeded but the database does not have it yet.

Write-behind is not a free performance upgrade. It is an asynchronous write pipeline, and it must be designed like one.

This chapter walks through how write-behind caching works, why it improves write latency, where data can be lost, batching and coalescing, ordering, retries, failure handling, when the pattern is safe to use, and what to monitor in production.