Deployment Options
Olympe has four modes of operations:
Community
This is the free version of Olympe. You don't have to setup anything: just open the Community and register for free.
Olympe Cloud
A dedicated instance running as SaaS and operated by Olympe. Please contact us for more information about how to setup a fully managed Olympe instance.
SLAs
When contracting with Olympe for a dedicated instance, you can expect the following SLAs: 99.7% uptime, working hours support, and a 24-hour response time for critical issues.
99.7% uptime means that the service can be down for a maximum of 0.3% of the time in a month, which is equivalent to 2 hours, 10 minutes and 24 seconds.
The instance is considered as being up and running according to the following criteria: all processes running on the instance are up and running, and the instance is reachable from the internet: the backends, the orchestrator delivering the code as data as well as the nginx (frontend) delivering frontend applications.
The instances are operated by Kubernetes which includes a built-in health check ability. Every 10 seconds, a check is asked to each resource to ensure it is still responding as well as correctly responding. The whole instance is considered as being down (not healthy) if at least 1 resource is down. A resource is considered as being down if the health check fails more than 3 times in a row.
The global uptime of the instance is calculated as the time the instance is healthy divided by the total time in a month. Metrics used for that result are generated every minute, based on the last Kubernetes health check.
Health checks executed by default for each deployed backend are the following:
- The backend is connected to the orchestrator
- The backend authentication state is correct (either authenticated or guest)
- An application has been started
- If relevant, the backend is connected to the bus: based on the application running, if the backend offers remote services accessible through the Olympe bus.
- Data sources run by the application are correctly connected
That list of checks can be extended with custom items if needed.
Olympe cannot be held responsible for downtime being caused by the custom checks.
Hybrid
The hybrid approach is Olympe Cloud with additional back-ends running elsewhere: on the edge, on premise or in a private cloud.
On Premise / Private Cloud
A full installation on your own infrastructure or a cloud provider of your choice. The best and easy way to install an on premise version of Olympe is by using Kubernetes and our Helm Chart.
For more information, please check the related documentation
Capabilities
The capabilities of the different options are as follows. The Hybrid option is omitted as it is really a mix of Olympe Cloud and On Premise: everything running within Olympe Cloud is treated according to Olympe Cloud rules, while the components running on your infrastructure are subject to your own guidelines and best practices, as per On Premise below.
| Community | Olympe Cloud | On Premise / Private Cloud | |
|---|---|---|---|
| SSO | N/A | Available | Available |
| Custom domain name | N/A | Available | Available |
| Logging | N/A | Available | Manage logs using your existing tooling and according to your current best practices. |
| Monitoring | N/A | Available | Monitor Olympe components using your existing monitoring tooling and best practices. |
| CI/CD Pipelines | N/A | Available | Implement deployment pipelines within your existing CI/CD infrastructure. |
| Code repository | N/A | Available. Builds are triggered from the Olympe git repository. Integrations with your repositories are possible. | Use your own repositories. |
| Configuration | N/A | Managed within a repository (GitOps) | Manage configurations in line with your current best practices. |
| Secrets management | N/A | Available | Use your existing secret management infrastructure. |
| Platform upgrades | N/A | Available | Can be supported by Olympe. |
| Artefact registry (NPM and Docker) | N/A | Available. Hybrid setups are possible. | You provide artefacts registry (if relevant). |
Migrating From One Mode To Another
From Community to Olympe Cloud and Hybrid
When moving from Community to Olympe Cloud, the following steps must be taken in consideration:
- Consider adding SSO (SAML or OAuth) for user authentication
- Projects initialized in the community must be imported into your dedicated environment
- For code to be deployed on the Olympe Cloud, it must be available from the Olympe CI/CD. The simplest mechanism is to implement a replication between your git repository to the Olympe git repository.
- The configuration of back-ends must be updated to point to the dedicated instance instead of the community. Also back-ends must be authenticated.
From Olympe Cloud / Hybrid to On Premise
On premise and private cloud deployments are typically performed by deploying Docker containers. These can be run both as a docker-compose for simpler scenarios, or more generally onto Kubernetes or OpenShift. Olympe provides a Helm chart for that purpose.