Persistent Storage

Persistent Storage allows you to mount S3-compatible buckets as persistent file storage for Edge Functions. Files survive function invocations and can be accessed via /s3/YOUR-BUCKET-NAME prefix, enabling up to 97% faster cold starts.

Key benefits

1. Dramatically faster cold starts: Up to 97% improvement in cold start times.
2. Persistent across invocations: Files survive between function executions.
3. S3 protocol compatibility: Mount any S3-compatible bucket, including Supabase Storage.
4. POSIX-like file system: Read and write files using standard Deno APIs.
5. Large file processing: Handle large files without fetching on each invocation.

Storage types

Persistent Storage backed by S3 protocol survives invocations. Ephemeral Storage (/tmp directory) resets on each invocation.

File operations supported

Deno.readFile(), Deno.writeFile(), Deno.readDir(), and other standard file operations.

Persistent Storage is valuable for:

• Large file processing
• Caching between invocations
• Custom image manipulation workflows
• Processing archives (zip files)
• Machine learning model storage

Setup

Requires S3 credentials as environment variables: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION, AWS_ENDPOINT_URL_S3.

Persistent Storage transforms Edge Functions into stateful, high-performance computing environments.