Which cache write policy can lose acknowledged writes if the cache node crashes?

Options

• ✗ Write-through: write the cache and the database before acknowledging
• ✓ Write-back: acknowledge after writing the cache, flush to the database later Correct
• ✗ Write-around: write only the database and let reads fill the cache
• ✗ Cache-aside with a TTL

Why

Write-back acknowledges before the data reaches durable storage, so anything not yet flushed dies with the node. It buys the lowest write latency at that price.

Review this in your daily deck → All cards in Caching