Skip to content

Storing data in the local filesystem

When to use a local volume inside a single container — keeping data through redeploys without sharing it across nodes.

Not every workload needs shared storage. When data must persist through the lifecycle of a single container — but does not need to be reachable from other nodes — a local volume inside that container is the simplest answer.

When local storage is the right pick

A local volume is appropriate when:

  • The data is private to that node — logs, scratch space, a single-node application’s working files.
  • You want the data to survive container redeploy. A redeploy rebuilds the container image but preserves declared local volumes.
  • You do not need to share the data with other layers or environments. If you ever need to, you can still re-mount the folder later.
  • Network availability of the data is not a concern — local volumes are always accessible because they sit on the same kernel as the process reading them.

A typical example is a load balancer’s local configuration files. You do not want the balancer to forget its upstream topology on a redeploy. Putting that config under a local volume keeps it across image refreshes without bringing in NFS or external storage.

When to choose something else

A local volume is the wrong tool when:

  • Multiple nodes in the same layer must read the same data. Use a mount onto a Shared Storage Container instead.
  • The data must survive total destruction of the container. A local volume disappears if you delete the node. Critical data should land on a dedicated storage container or an external NFS server.
  • You want the data to follow the application as it moves regions or accounts. Local volumes do not travel through environment export.

Behaviour notes

  • Local volumes are declared in the container’s volumeMounts / image configuration. The platform persists declared paths across redeploy operations.
  • Quotas on local volumes count against the container’s disk limit.
  • Files in a local volume are not part of any built-in snapshot or backup mechanism. If they matter, either mirror them to a Shared Storage Container on a schedule, or use the database backup add-on if the data is a database.

Where to go from here

  • Mount points — how to attach a remote folder if local storage is not enough.
  • Storage exports — exporting data from one node to another in the same account.
  • Dedicated storage container — the shared option, when local is too narrow.