Production Data and Storage

Build the mental model

Every production app has three very different kinds of data, and treating them the same is one of the most common ways apps break after launch.

• Structured data (users, orders, comments) belongs in a database — transactions, indexes, consistent queries.
• Files (images, videos, PDFs, reports) belong in object storage — built for large binary blobs served over HTTP.
• Object storage is not a database, and despite appearances, it is not a folder on your server either.

A production database connection also carries higher stakes than a local one: real credentials, real customer data, network rules, and backup schedules all matter now.

Decide before writing upload code: database record, object storage file, or genuinely disposable data. Getting this wrong rarely shows up in development — it shows up in production, usually as a support ticket.

Object Storage

A system built to store large binary files (images, videos, backups) cheaply and serve them over HTTP, indexed by key rather than by folder path.

Ephemeral Filesystem

Local disk storage that can be wiped when a container restarts, redeploys, or moves to a new machine.

Persistent Storage

Storage that lives outside the app's own container and survives restarts and redeploys, such as a managed database or object storage bucket.

PRODUCTION DATA PATHS
---------------------
PRODUCTION DATA PATHS
----------------------

           [ APP SERVER ]
             |        |
      writes |        | writes
             v        v
      [DATABASE]  [OBJECT STORAGE]
      rows/tables   files/images/videos
      (structured)  (blobs)


  LOCAL DISK (container)      MANAGED STORAGE
  -----------------------     -----------------------
  EPHEMERAL                   PERSISTENT
  wiped on redeploy/restart   survives redeploy/restart

Connect it to a real scenario

A tiny classifier makes the decision concrete: given a piece of data, decide which of three destinations it belongs to.

Ask: would losing this on the next deploy be a disaster, an inconvenience, or a non-event? That answer almost always resolves the ambiguity.

• Database, object storage, and ephemeral storage each solve a different problem — pick per data type, not per convenience.
• Local disk on most production hosts is ephemeral: files saved there can vanish on the next deploy or restart.
• Decide where new data lives before writing upload code, not after a user reports a missing file.

Try the working example

function classifyStorage(type) {
  const objectStorageTypes = ["image", "video", "document", "backup"];
  const databaseTypes = ["user-profile", "order", "comment"];
  const ephemeralTypes = ["cache-value", "session-token"];

  if (objectStorageTypes.includes(type)) return "object-storage";
  if (databaseTypes.includes(type)) return "database";
  if (ephemeralTypes.includes(type)) return "ephemeral (ok to lose)";
  return "unknown";
}

const items = [
  { type: "image", label: "user avatar upload" },
  { type: "user-profile", label: "username + email row" },
  { type: "cache-value", label: "rate-limit counter" },
  { type: "backup", label: "nightly database dump" },
];

for (const item of items) {
  console.log(`${item.label.padEnd(28)} -> ${classifyStorage(item.type)}`);
}

user avatar upload           -> object-storage
username + email row         -> database
rate-limit counter           -> ephemeral (ok to lose)
nightly database dump        -> object-storage

MDN — HTTP and file uploads — Cloud & Deployment