Streaming and sync

File streaming versus on-demand sync: what should you actually compare?

See how Suite file streaming supports large shared libraries, familiar applications and focused local storage compared with sync workflows.

Reach shared files without preparing a full local library

Reach shared files without preparing a full local library

File streaming shines when people need familiar file paths into a large shared library while working on only part of it. Suite brings those files into a mounted drive and streams requested data, helping reduce full-copy preparation and focus local storage on active work. Compare that experience with the specific sync product you use, including any on-demand features, to see which approach best supports your applications and handoffs.

File streaming shines when people need familiar file paths into a large shared library while working on only part of it. Suite brings those files into a mounted drive and streams requested data, helping reduce full-copy preparation and focus local storage on active work. Compare that experience with the specific sync product you use, including any on-demand features, to see which approach best supports your applications and handoffs.

Suite presents shared storage as a mounted filesystem. This is valuable for applications that expect files and folders, especially when preparing local replicas delays work. Plan around the application’s read pattern, connection and cache to make the most of the workflow. See our filesystem overview.

Suite presents shared storage as a mounted filesystem. This is valuable for applications that expect files and folders, especially when preparing local replicas delays work. Plan around the application’s read pattern, connection and cache to make the most of the workflow. See our filesystem overview.

Account for what modern sync products already do

Is file streaming the same as online-only sync?

Microsoft’s OneDrive Files On-Demand supports online-only files and files kept locally for offline use. Opening an online-only file downloads it; a file marked to remain on the device consumes local space and can be used offline. A fair comparison must include those modes rather than assume every sync product copies the entire library to every computer. See Microsoft’s explanation of Files On-Demand.

No. Both can make a large cloud library visible without downloading it all first. The useful distinction is how the selected file reaches the application when work begins. Compare these three behaviors:

  • Keep files locally: Selected files are downloaded and retained on the device. This supports offline access where the product provides it, with local capacity reserved for those files.

  • Download when opened: An online-only entry becomes a local file on use. Microsoft’s OneDrive guidance documents this behavior and the option to keep files on the device. Online-only availability already avoids a full-library download.

  • Stream the requested file data: Suite delivers the portions the application asks for through a mounted filesystem. Backblaze’s Suite announcement describes this for B2: the application can use file data without first preparing the entire dataset locally. That makes Suite worth evaluating for active work with large shared media, beyond browsing an archive.

The purchasing question becomes more precise: which access model serves this workload with acceptable waiting, storage use and operational effort?

Product labels alone do not establish identical behavior. Compare the actual file-open and editing task, with the same source media and initial conditions.

Can I edit through Suite without pre-caching or making proxies first?

Identify the work that could benefit

Yes. Suite supports direct streaming, and its enterprise overview describes working with high-resolution media without waiting for downloads, proxies or syncing. Deliberate pre-caching and proxy creation are optional workflow choices, not prerequisites for access. Automatic client caching and the editing application’s generated caches are separate.

Choose the workflow around your footage, connection and workstation. Follow the Premiere Pro shared-media guide for an editing example. Streaming supplies file access; the editor’s application still manages project collaboration and media processing.

Identify the work that could benefit

Consider a hypothetical editor who moves among projects in a shared library. The editor may need to browse broadly, inspect a few assets and then concentrate on a smaller project. Requiring a full local replica before every switch could create unnecessary preparation. A streaming filesystem is a candidate for reducing that preparation.

Now consider a processing job that reads every byte of every input. It still needs all those input bytes delivered somewhere the computation can use them. Streaming may change the staging process, but it does not remove that data requirement. Measure the complete job rather than celebrate the moment its first file opens.

These are workload examples, not measured Suite customer results.

Use five comparison questions

  1. When does useful work begin? Define an action such as displaying a required frame or opening a usable project. Measure that action rather than the appearance of a filename.

  2. What does the application actually read? Include background analysis, indexing and preview generation. Those activities may fetch more data than a person explicitly opens.

  3. What remains on the device? Measure local disk use after the same work, including product caches, application caches and generated outputs.

  4. What happens when the connection disappears? Evaluate the documented offline behavior for the chosen configuration. Suite requires an internet connection; pre-caching is not an offline mode. See our connectivity requirement.

  5. How is a handoff verified? Finish a test write, confirm upload completion, and have the intended collaborator open the output. An operating-system copy dialog is not sufficient evidence that Suite has finished uploading. See our upload guidance.

Keep first use and repeat use separate

Run the same task with clearly recorded initial conditions. A test with already cached data answers a different question from a first access to new material. Neither is inherently unfair; presenting them as interchangeable is.

Record the host, network, application version, project, local capacity and storage location. Report errors and waiting as well as successful runs. If one product requires a preparatory download, count that preparation when it is part of the normal workflow.

Choose the next step from the result

Evaluate Suite when local-copy preparation or frequent project changes are a meaningful obstacle and connected filesystem access fits the work. Preserve a supported offline workflow when disconnection is essential. Keep an existing approach when a fair pilot does not show enough improvement to justify changing it.

Use the Managed or S3 Native paths below to discuss the appropriate deployment. This comparison provides an evaluation framework; it does not claim a universal speed advantage or report a completed benchmark.

Consider a hypothetical editor who moves among projects in a shared library. The editor may need to browse broadly, inspect a few assets and then concentrate on a smaller project. Requiring a full local replica before every switch could create unnecessary preparation. A streaming filesystem is a candidate for reducing that preparation.

Now consider a processing job that reads every byte of every input. It still needs all those input bytes delivered somewhere the computation can use them. Streaming may change the staging process, but it does not remove that data requirement. Measure the complete job rather than celebrate the moment its first file opens.

These are workload examples, not measured Suite customer results.

Why the amount read matters more than the file size

Imagine a 100 GB source file and an application that requests 2 GB of distinct data while inspecting selected shots. A workflow that must first download the entire file has a 100 GB preparation step. A streaming workflow can deliver the requested portions as the application asks for them, allowing useful work to begin before all 100 GB arrives. The opportunity is avoiding unnecessary preparation, not making the network infinitely fast.

This is a simplified illustration, not a Suite benchmark or a prediction that exactly 2 GB crosses the network. Read-ahead, repeated reads, thumbnails, indexing and other application activity can change the bytes transferred. A full-file export or analysis pass may eventually request the entire source. Modern on-demand sync also avoids downloading an entire library in advance; compare what happens when the selected large file is opened, not just whether its name appears.

For a useful pilot, record three separate quantities: source-file size, observed network transfer, and time to the first usable application result. That connects the benefit to the actual job. Use the full-input workflow guide when processing needs every byte.

Use five comparison questions

  1. When does useful work begin? Define an action such as displaying a required frame or opening a usable project. Measure that action rather than the appearance of a filename.

  2. What does the application actually read? Include background analysis, indexing and preview generation. Those activities may fetch more data than a person explicitly opens.

  3. What remains on the device? Measure local disk use after the same work, including product caches, application caches and generated outputs.

  4. What happens when the connection disappears? Evaluate the documented offline behavior for the chosen configuration. Suite requires an internet connection; pre-caching is not an offline mode. See our connectivity requirement.

  5. How is a handoff verified? Finish a test write, confirm upload completion, and have the intended collaborator open the output. An operating-system copy dialog is not sufficient evidence that Suite has finished uploading. See our upload guidance.

Keep first use and repeat use separate

Run the same task with clearly recorded initial conditions. A test with already cached data answers a different question from a first access to new material. Neither is inherently unfair; presenting them as interchangeable is.

Record the host, network, application version, project, local capacity and storage location. Report errors and waiting as well as successful runs. If one product requires a preparatory download, count that preparation when it is part of the normal workflow.

Choose the next step from the result

Evaluate Suite when local-copy preparation or frequent project changes are a meaningful obstacle and connected filesystem access fits the work. Preserve a supported offline workflow when disconnection is essential. Keep an existing approach when a fair pilot does not show enough improvement to justify changing it.

Use the Managed or S3 Native paths below to discuss the appropriate deployment. This comparison provides an evaluation framework; it does not claim a universal speed advantage or report a completed benchmark.

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.