Storage

As described on our Architecture page, Vault's storage backend is untrusted storage used purely to keep encrypted information.

Supported storage backends

For enterprise customers, HashiCorp provides official support for Vault's Integrated Storage and Consul as storage backends. Vault Enterprise customers are strongly recommended to use these supported storage backends for best outcomes. Version 1.12.0 of Vault Enterprise will not start if configured to use a storage backend other than Integrated Storage or Consul. This was meant to protect against issues caused by using unsupported backends that do not support transactional storage. Version 1.12.2 modified this behavior to instead log a warning when unsupported storage backends are used, while ensuring that Vault will start.

Many other options for storage are available with community support for Vault - see our Storage Configuration section for more information.

Choosing a storage backend: Refer to the integrated storage vs. external storage section of the storage configuration page to help make a decision about which storage backend to use.

Backups

Due to the highly flexible nature of Vault's potential storage configurations, providing exact guidance on backing up Vault is challenging.

When backing up Vault, there are two pieces to consider:

Vault's encrypted data in the storage backend
Configuration files and management scripts for running the Vault server

There's also a big question - what is the error case you're trying to guard against by saving a backup?

The big question - why take backups?

It's important to consider the question of "why take a backup" while developing your ongoing backup and disaster recovery strategy.

Taking a backup is recommended prior to upgrades, as downgrading Vault storage is not always possible. Generally, a backup is recommended any time a major change is planned for a cluster.

More specifically, we recommend taking backups before, but not during, write operations to the /sys API (excluding the /sys/leases, /sys/namespaces, /sys/tools, /sys/wrapping, /sys/policies, and /sys/pprof endpoints). Some examples of workflows that write to the /sys API are upgrades and rekeys. In the future, this guidance may change for the Integrated Storage backend.

Backups can also help with accidental data deletions or modifications. In this case, the story can get a little tricky. If you simply recover a backup from 5AM with the correct data, but the current time is 10AM, you will lose data written between 5 and 10AM. Lucy Davinhart gave a HashiConf talk that serves as an interesting case study.

We do not recommend backups as protection against the failure of an individual machine. Vault servers can run in clusters, so to protect against server failure, we recommend running Vault in HA mode. With community features, a Vault cluster can extend across multiple availability zones within a region.

Vault Enterprise supports replicated clusters and disaster recovery for data center failure. When using Vault Community Edition in HA Mode, a backup can help guard against the failure of a data center.

Ultimately, backups are not a replacement for running in HA, or for using replication with Vault Enterprise. As you develop a plan for recovering from or guarding against failure, you should consider both backups and HA as critical components of that plan.

Backing up vault's persisted data

Backups and restores are ideally performed while Vault is offline. If offline backups are not feasible, we recommend using a storage backend that supports atomic snapshots (such as Consul or Integrated Storage).

If your storage backend does not support atomic snapshots, we recommend only taking offline backups.

To perform a backup or restore of Vault's encrypted data when using a HashiCorp-supported storage backend, see the instructions linked below. For other storage backends, follow the documentation of that backend for taking and restoring backups.

Integrated Storage snapshots
Consul snapshots

Backing up multiple clusters

If you are using Vault Enterprise Performance Replication, you should plan to take backups of the active node on each of your clusters.

Configuration

In addition to backing up Vault's encrypted data via the storage backend, you may also wish to save the server configuration files, any scripts for managing the Vault service, and ensure you can reinstall any user-installed plugins. The location of these files will be specific to your installation of Vault.

NOTE: Although a backup or snapshot of Vault's data from the storage backend is encrypted, some of your configuration may be sensitive (a Vault token for Transit Autounseal or a TLS private key in your configuration, for example). The presence of this information in your backups will mean that they may need to be carefully protected.