• Document Under Review
  • Since 4.0.0

Availability

CrafterCMS supports geo-distributed deployments with multiple data-centers per geography making the uptime of the delivery tier near 100%. Traffic to the content delivery system flows to the nearest healthy data-center providing speed and availability.

Availability can be divided into High-Availability (HA) and Disaster Recovery (DR). Since CrafterCMS is two separate subsystems, delivery and authoring, then will describe HA and DR in the context of content delivery and then in the context of content authoring.

Delivery

Content Delivery in CrafterCMS is based on the principals of Shared Nothing Architecture. This means delivery nodes don’t intercommunicate, are not backed by a single database nor file system. As such, adding new nodes increases reliability since the failure of one node doesn’t result in a failure in content delivery.

Ideally, an installation would follow an N+1 delivery allocation, where N is the number of nodes required to handle the delivery load and the extra node is in case of a single node failure.

Distributing nodes across geographies (multi-region) means automatic Disaster Recovery is built-in. A failure of a data center or several data centers in a region will not bring down content delivery.

Authoring

Content Authoring is the tier used by the few authors to generate content (XML, code, and static assets) to be published to the delivery tier. As such, there are much less stringent availability requirements for the authoring tier.

The authoring application has built-in primary/replica clustering. Being a stateful application, the clustering is based on replicas following the primary with a fronting load-balancer that directs traffic to the primary while the primary node is healthy. Fail-over is automatic with the replica becoming a primary and switching its health status so the load-balancer directs traffic to it instead of the defunct primary.

More information on clustering here