Backup and Restore Architecture

Ataccama ONE PaaS is created in accordance with infrastructure as code principles and techniques. This means that the whole Ataccama ONE PaaS solution, including Ataccama ONE itself, is deployed based on code and application images stored and maintained in the Ataccama cloud GitLab.

Thanks to this architecture, backing up the whole infrastructure in practice corresponds to backing up the Ataccama cloud GitLab resources. The following diagram shows the backup service design of Ataccama ONE PaaS:

Ataccama ONE PaaS backup parameters are optimized to balance data security and efficiency of the backup process.

The parameters are described in the following table:

Customer data backup is an integral part of the Ataccama ONE PaaS service that is fully automated and preconfigured, with no customer action required. For disaster recovery purposes, backup snapshots are stored in an S3 bucket. At rest, backups are encrypted using AES-256 encryption.

The Amazon S3 service ensures durability of stored objects by replicating them across multiple facilities located in the same regulatory region.

Ataccama ONE PaaS uses two different backup techniques:

The following diagrams provide an overview of the backup service design in Ataccama ONE PaaS deployed in an AWS or Azure environment:

Customer data backup parameters are optimized to balance data security and efficiency of the backup process as well as the amount of storage needed to hold backed up snapshots.

The parameters are described in the following table:

As mentioned previously, Ataccama ONE PaaS is created in accordance with infrastructure as code principles and techniques, meaning that the whole Ataccama ONE PaaS solution, including Ataccama ONE itself, is deployed based on code and application images stored and maintained in the Ataccama cloud GitLab.

Thanks to this, restoring the platform and its infrastructure corresponds in practice to a fresh installation from Ataccama cloud GitLab resources using Terraform, Helm charts, and other automation tools.

The following diagram shows the restore service design of Ataccama ONE PaaS:

The customer data restore process uses the same mechanisms and services as the backup service. Restore mechanisms are an integral, highly automated part of Ataccama ONE PaaS. The restore process is initiated and managed by the Ataccama Operations team following a disaster situation.

For more information about restore scenarios for disaster situations, see Disaster recovery.

Ataccama ONE PaaS uses two different restore techniques, in analogy to backup methods:

The following diagrams provide an overview of the restore service design in Ataccama ONE PaaS deployed in an AWS or Azure environment:

During usual system operation, files produced by Ataccama ONE are stored to an external directly mounted filesystem provided as a service by the cloud provider. Corruption or loss of data can occur due to software bugs, human error, or storage service problems.

Once the Ataccama Operations team detects the issue, they inform the customer and perform a semi-automated restore, which means that files are restored to their last backed up state (unless agreed otherwise).

To ensure the consistency of the Ataccama ONE Platform data set, files must also be restored on the filesystem used by Ataccama ONE PaaS. The specifics of this procedure are then discussed with the customer, including the exact recovery point that will be restored.

The following restore service parameters apply in this scenario:

*Restoring only a part of the environment requires a hands-on course of action and coordination with data stewards. Therefore, RTO depends on the specific situation and its complexity.

During usual system operation, data produced by Ataccama ONE is stored to the database provided as a service by the cloud provider. Corruption or loss of data can occur due to software bugs, human error, or storage service problems.

Once the Ataccama Operations team detects the issue, they inform the customer and perform a semi-automated restore, which means that the data is restored to its last backed up state (unless agreed otherwise).

To ensure the consistency of the Ataccama ONE Platform dataset, the data stored in the database within Ataccama ONE PaaS must also be restored. The specifics of this procedure are then discussed with the customer, including the exact recovery point that will be restored.

The following restore service parameters apply in this scenario:

*Restoring only a part of the environment requires a hands-on course of action and coordination with data stewards. Therefore, RTO depends on the specific situation and its complexity.

A substantial part or the whole environment of Ataccama ONE PaaS is corrupted, lost, or misconfigured to the point that it cannot be operated without restore. Corruption or loss of the environment can occur due to human error, serious connectivity problems, cloud service provider malfunctions, or software bugs.

Once the Ataccama Operations team detects the issue, they inform the customer and restore the service in the same or different cloud provider region depending on the mutual agreement. The specific location is selected based on the disaster situation and must be approved by the customer.

In case the environment needs to be completely restored, a fresh installation is performed instead while customer data and configurations are retrieved using automation tools. All customer data is restored to its latest backed up state unless agreed otherwise.

The following restore service parameters apply in this scenario: