Atlassian's Cloud Backup Policy

What is Atlassian’s Cloud Backup Policy? This is one of the most repeated questions in conversations about Jira Cloud.

Atlassian does indeed have a backups policy in Cloud, but from our point of view, with some limitations.

As a general rule, backups of the Jira Cloud instance are expected to be performed by org admin or site admin users manually. When a change is planned that may have a major impact on the instance, a manual backup should be performed in order to restore the instance as it was, to rollback if necessary.

But what if you don’t have that backup before you make those massive changes, is there a way to restore the instance back to the way it was?

Here is a compilation of the most frequently asked questions and answers on this topic:

Does Atlassian perform regular backups of Cloud instances?

Yes, Atlassian has a backups policy for internal emergency recovery purposes. It is important to clarify that these backups are different from data backup that we can generate manually as administrators. For starters, they are not intended to recover data deleted by users, but are designed for disaster recovery at the infrastructure level. The ultimate responsibility for backing up an instance’s data rests entirely with the organisation and its administrators.

More information provided by Atlassian: Can Atlassian RDS backups be used to revert changes?

“We cannot use our RDS backups to revert changes. These include changes such as fields overwritten by scripts or issues, deleted projects or sites.

This is because our data is not stored in a single central database. Instead, it is stored in many microservices, which makes reverting changes a risky process.

To avoid data loss, we recommend making regular backups.”

To avoid data loss, we strongly recommend regular backups of the instance.

How often are Atlassian backups performed and where are they stored?

Automated Amazon Relational Database Service (RDS) backups are generated daily and retained for 30 days to allow data to be restored at a moment’s notice. RDS snapshots are encrypted at rest.

When can Atlassian be asked to restore the instance from its backup?

Under normal operating circumstances, Atlassian does not restore instance data, but whenever an out-of-the-ordinary situation arises, such as an unlikely failure in Atlassian’s infrastructure or any issue that prevents the instance from functioning as expected due to internal services or deployments, it is possible to request a restore of the site.

Most of the time this will be done by Atlassian proactively, but if this does not happen, Atlassian will evaluate the request and, if approved, restore the instance to a specific point in time.

It is not possible to recover/unwind only specific data, the entire site is restored.

Do the backup that Atlassian performs include apps data and automation data? What data would not be recovered?

Apps data is stored on their respective vendor’s servers, so there is no guarantee what would happen to third party data. This would need to be checked with the provider of each application.

Generally speaking, a restore by Atlassian returns the instance to a certain point in time, in all respects, except for additional third party data.

In the event that we have a planned intervention that affects the entire instance, can we contact Atlassian for a backup?

Atlassian can be contacted, but they will tell us to do a manual backup to ensure that all data on the instance is safe. Instance administrators are expected to perform these backups periodically.

What is the recommendation if a mass deletion of data cannot be foreseen and a restore is needed?

If the deletion is related to normal user operation, using normal Jira functionality and no system error has occurred, then we will need to use any previous backups we have generated. Atlassian will not be responsible for the recovery of such data.

Ideally, having a good backups policy will help to avoid such problems in the instance.