> For the complete documentation index, see [llms.txt](https://wiki.datadike.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://wiki.datadike.com/pam/deploy-operate/availability-and-sla.md).

# Availability and continuity

Availability levels, redundancy, continuity objectives and the update model, for the SaaS and on-premises delivery modes.

## What this is and when to use it

DataDike PAM is delivered either as **SaaS**, on the DataDike cloud, or **on-premises**, in the customer's own data centre. Both are designed for high availability, so that losing one component does not interrupt privileged access.

This page states the published availability terms and continuity objectives for each mode. Use it when the question is what is guaranteed rather than how it is built: a security questionnaire, a risk review, a continuity plan, or simply what happens when a node dies.

For how a cluster is actually assembled, see [Install a high-availability cluster](https://wiki.datadike.com/pam/deploy-operate/install-cluster.md). For what the architecture does with a failure, see [Architecture and components](https://wiki.datadike.com/pam/get-started/architecture.md).

{% hint style="warning" %}
The figures below are the published terms. The SLA, the support window and the hosting region that apply to a given deployment are the ones stated in its contract.
{% endhint %}

## SaaS

Delivered as Software-as-a-Service, hosting, infrastructure and operation are DataDike's responsibility, under these terms:

| Term             | What is published                                                                                                                                                     |
| ---------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Availability SLA | **99.9% minimum**, measured over the contract period                                                                                                                  |
| Licensing        | By subscription, including **continuous version updates** — fixes and new features at no extra cost while the subscription runs                                       |
| Hosting          | **In the country closest to the customer's geolocation where DataDike offers infrastructure services**, with an external replica placed by lowest-latency geolocation |

The support window, the severity levels and the response times are published separately, in [SLA and support](broken://pages/FltvOrzlfOQ2yhTKmvCz).

### Redundant local connector

The SaaS offer includes a **redundant local connector**, so that access keeps working if the link to the cloud drops, without users or components noticing.

* **Automatic, real-time synchronisation between components.** If one site's components become unavailable, the other site's components take over every service with no data loss and no performance degradation.
* **One-way connection.** Connections are only ever made from the customer environment towards DataDike's cloud, never the reverse. This matters for a firewall review: no inbound rule from the internet is required.

## On-premises

Installed in the customer's data centre, the solution runs **active/active**.

Additional items needed to make that work are supplied according to the sizing — database or Windows RDS licensing and HA licences, for example. The list for a given deployment comes from DataDike, not from this page; see the gaps table in [System requirements](https://wiki.datadike.com/pam/deploy-operate/requirements.md).

## Continuity and recovery

Both modes share the same continuity design:

| Objective    | Value            |
| ------------ | ---------------- |
| Architecture | Active/active    |
| **RTO**      | Under 30 seconds |
| **RPO**      | Under 1 second   |

These figures assume the active/active reference architecture as sized by DataDike for the deployment, including the database and storage replication it specifies. On-premises they hold to the extent that the infrastructure meets that sizing, so confirm the target for your own deployment before committing to these numbers in a continuity plan.

* **Backups** carry no retention restriction over the years, and are encrypted and protected against tampering.
* **Component redundancy** removes the single point of failure from the vault, the session-brokering proxy and the administration console.

{% hint style="danger" %}
High availability is not a backup, and the RPO above is not a retention policy. Replication propagates a deletion exactly as fast as it propagates a write. See [Back up and restore](https://wiki.datadike.com/pam/deploy-operate/backup-and-restore.md).
{% endhint %}

## Separation between customer environments

Different customers' environments operate **logically segregated and independent**: consoles, vaults and audit trails with no relationship between them.

## How to verify it worked

Availability terms are only worth what a test proves.

1. **Fail a node on purpose**, in a maintenance window, and measure how long access actually takes to return. That measurement is the recovery time of your own deployment.
2. Confirm what the failover cost you: check whether sessions open at the moment of the failure survived, and whether the audit trail recorded them continuously.
3. For SaaS, test the redundant local connector by dropping the link to the cloud, not by reading about it.
4. Restore a backup into a spare environment and time it. An untested backup is an assumption.

Record the date and the measured figures. A continuity review is answered by the last test, not by the design objective.

## Related pages

* [SLA and support](broken://pages/FltvOrzlfOQ2yhTKmvCz) — the availability figure, the severity levels and the response times
* [Install a high-availability cluster](https://wiki.datadike.com/pam/deploy-operate/install-cluster.md) — how the active/active arrangement is built
* [Back up and restore](https://wiki.datadike.com/pam/deploy-operate/backup-and-restore.md) — what high availability does not cover
* [System requirements](https://wiki.datadike.com/pam/deploy-operate/requirements.md) — sizing, and the licensing that on-premises HA adds
* [Monitoring and health](https://wiki.datadike.com/pam/deploy-operate/monitoring.md) — watching an instance so you learn about a failure before your users do
* [Architecture and components](https://wiki.datadike.com/pam/get-started/architecture.md) — what each redundant component does
