ARTICLE · Systems
Disk and memory
A value can exist in your program without being safely saved. Memory and persistent storage have different lifetimes, and users care about the difference.
Disk and memory
YouTube stays disconnected until you choose to load this player. Loading connects to Google/YouTube, which receives your IP address and may use cookies or browser storage.
You type a character biography and see it on screen. That proves the interface can display your draft. It does not yet prove the draft will survive a refresh, a process restart, or a machine failure.
Memory holds the working state a running program uses. Persistent storage is intended to retain data beyond that program's lifetime. A database builds additional behavior around storage, such as queries, indexes, and rules for coordinating updates. These are related layers rather than interchangeable names for “where data goes.”
Ask what survives the next interruption#
A JavaScript array disappears when its process exits. A file on a persistent disk may survive that restart, but a file inside a disposable deployment may not survive replacement. A database write has its own durability guarantees and failure conditions.
That is why the reload test is so revealing. Create a record, reload, and trace where the second request obtains it. If the result survives only because the frontend still remembers it, you have demonstrated a convincing screen rather than persistence.
The Vault gives the course a place for records to outlive the server process. Keep the episode's storage imagery, then attach a lifetime to every box in your diagram: this request, this process, this deployment, or the durable record the user expects to find tomorrow.
See also