Why should an application typically avoid storing uploaded files directly on the same server instance that runs the web application?

Storing uploaded files on local disk ties that data to a specific server instance, which breaks horizontal scaling (a file uploaded to one instance wouldn't be visible from requests handled by a different instance) and risks permanent data loss if that specific instance is terminated (common in cloud environments with ephemeral, disposable instances) -- the standard solution is storing uploaded files in a dedicated object storage service like Amazon S3, which is durable, accessible from any application instance, and decoupled entirely from the application server's own lifecycle.

// Fragile: tied to this specific server's local disk
await fs.promises.writeFile(`./uploads/${filename}`, fileBuffer);

// Durable and scalable: stored in shared cloud object storage
await s3Client.send(new PutObjectCommand({ Bucket: 'uploads', Key: filename, Body: fileBuffer }));