A scaled back end deployment is the most common starting point for high availability, with regard to |chef server oec|. It is easier to scale out and load balance the front end servers later, but it's essential to set up the back end servers for primary/secondary failover, system backups, and so on, as part of the initial configuration. Most |chef server oec| customers choose this configuration.
where:
- The front end server is shown as a single machine. Requests to the front end are done via the |api chef server| and are made by the |chef client|, |knife|, and users (when they log on to the |webui| web user interface). The need for load balancing the front end servers depends on the number of requests that will be made to the |chef server| and on the capabilities of the hardware on which the front end services are installed. In many organizations, a single front end machine is fine and the decision to add a machine (and add load balancing) can be made later and with minimal disruption.
- The back end servers handle data storage and retrieval. Failover is achieved on the back end servers using a combination of asynchronous block level replication of logical volume managers (|drbd|), a shared virtual IP address, and access to the primary server (via the shared virtual IP address) maintained by |keepalived|.