Chapter 5. 3scale API Management High Availability and Evaluation
- You need to have an available OpenShift cluster to deploy elements of the High Availability and Evaluation templates.
5.3. High Availability template
The High Availability (HA) template allows you to have a HA setting for critical databases.
- Before deploying the HA template, you must deploy and configure the external databases, and configure them in a HA configuration with a load-balanced endpoint.
5.3.2. Using the HA template
For HA, the template named
amp-ha-tech-preview.yml allows you to deploy critical databases externally to OpenShift. This excludes:
Differences between the standard
amp.yml template and
Removal of the following elements:
- backend-redis and its related components
- system-redis and its related components
- system-mysql and its related components
- Redis and MySQL related ConfigMaps
- MYSQL_IMAGE, REDIS_IMAGE, MYSQL_USER, MYSQL_ROOT_PASSWORD parameters
By default, increased from 1 to 2 the number of replicas for non-database
Addition of the following mandatory parameters, allowing you the control of the location of external databases:
amp-ha-tech-preview.yml, you need to configure database connections (excluding
system-sphinx that do not contain permanent data) out of the cluster via the newly added mandatory parameters. The endpoints require database load-balanced connection strings, including authentication information. Also, for the non-database deployments, the number of pod replicas is increased to 2 by default to have redundancy at application-level.
5.4. Evaluation template
For evaluation purposes, there is a template named
amp-eval-tech-preview.yml that deploys a 3scale environment without resource requests nor limits.
The only functional difference compared to the standard
amp.yml template is that the resource limits and requests have been removed. This means that in this version the minimum hardware requirements have been removed on the pods at CPU and Memory level. This template is intended only for evaluation, testing, and development purposes as it tries to deploy the components in a best-effort way with the given hardware resources.