File streaming

When does an application need filesystem paths instead of object access?

Use Suite’s filesystem for familiar applications and S3 Native object access for compatible services in one shared workflow.

Bring the right interface to every part of the workflow

Suite S3 Native connects two useful ways of working with the same supported storage. People and file-based applications use a mounted filesystem; compatible services access native objects through the bucket’s API. Match each task to the interface its tools use, and you can bring existing assets into more workflows without redesigning every application around an object API.

For a workflow that needs both interfaces, S3 Native lets you work through filesystem paths while reading and writing standard objects in your existing storage. That makes an existing library useful to additional file-based workflows. Confirm the specific provider and application behavior before deployment. See our S3 Native overview.

Connect object keys to the paths applications use

Amazon S3 identifies each object by its key. Its ordinary bucket model is flat; prefixes and delimiters can present a logical folder structure. A key such as projects/demo/input.mov therefore does not, by itself, establish a local filesystem path that a desktop application can open. AWS explains this distinction in its object naming documentation.

Do not decide from the appearance of a slash in a name. Inspect the application's supported input interface. A field accepting a local path, a URL and an SDK call accepting a bucket/key pair represent different integration choices. Some applications support more than one; use their documentation and a representative task to choose.

Build a three-line interface inventory

Before changing storage, write down these three entries for each step of the workflow:

  • Input: which tool reads which asset, through which interface?

  • Output: where does that tool write its result, and who consumes it next?

  • Handoff: what observable event tells the next step that the result is ready?

This small inventory often reveals the actual obstacle. The problem may be a tool that only accepts paths, repeated preparation of local copies, or uncertainty about when another system can read a result. Each calls for a different evaluation. Adding a filesystem without identifying the obstacle makes success difficult to judge.

Work through a mixed example

Consider a hypothetical team with an object-aware catalog and a separate analysis tool that expects file paths. The catalog identifies the asset; the analysis tool reads it and creates a report; another service consumes the report.

A useful pilot would keep the catalog's existing access pattern and test whether the analysis step can use a supported Suite host. Record the path mapping, permitted input, output location and completion check. Then verify that the next consumer sees the intended result. Confirm compatibility with the particular catalog and analysis tools you choose.

Include failure cases in the pilot: a missing input, denied access and an interrupted operation. The agent's report should distinguish “the tool finished” from “the downstream consumer verified the output.” Define ownership of writes before testing competing updates to the same asset.

Make the handoff explicit with an asset manifest

For a mixed editor-and-agent workflow, record the asset’s logical identity separately from its access address. A useful manifest can contain the project name, relative asset path, input revision, interface used by each worker, and expected output location. This gives both sides a shared description without pretending that an S3 address is a local filename.

For example, an illustrative asset might have the relative path Project-A/Media/Interview-01.mov. An editor accesses it under the Suite drive configured on that workstation. An S3-capable service accesses the corresponding object through an administrator-confirmed bucket and key. These are examples of addressing, not prescribed mount locations or a guarantee that every existing bucket layout maps identically. Verify the actual mapping during setup.

When the service returns a transcript, include the source identity and revision beside the transcript path. The editor can then tell which media the output describes. If either side renames or replaces the source, update the manifest and recheck visibility before continuing. Keep access secrets out of the manifest; it should describe the work, not grant permission. The editor-and-agent workflow explains the wider handoff.

Choose a supported next step

If every step already uses object APIs successfully, identify a concrete benefit before introducing file access. If a path-based step repeatedly requires preparation, evaluate Suite against that step's real workload.

For existing customer-owned storage, BYO/S3 Native requires at least 20 TB of active storage, including agent customers. The BYO guide documents the minimum and assisted setup. Below that threshold, evaluate whether Suite-managed storage fits instead. A successful interface choice still needs supported hosts, permissions and an agreed completion check; it does not establish automated account provisioning.

Put the idea to work

Start with your workflow.

Bring shared files into your familiar applications with Suite Managed, or connect your own supported storage with S3 Native. S3 Native requires at least 20 TB of active storage.