High availability (HA) options for PI AF

HA Option

Pros

Cons

Failover Cluster

• Protection against front-end failures

• Requires significant initial investment in cluster hardware (e.g. shared storage)

• Shared storage limits the distance between cluster nodes

Network Load Balancing (NLB)

• Protection against front-end failures

• Allows for division of labor across multiple independent front ends

• Load balancer can be single point of failure

• If load balancer is not application aware, it may require additional configuration to detect certain types of front-end failures (e.g. service stopped)

PI AF Server Collective

Note: Supported in PI AF 2.1 and later. PI AF collectives are not recommended.

• Possible to direct read-only applications to secondary server(s), resulting in some division of labor

• Incomplete protection against front-end failures

  • Since PI AF secondary members are read-only, this option is not recommended for environments that need write access protection (asset analytics, notifications, event frames, etc.)

• Difficult to manage (for example, no ability to promote secondary server if primary unavailable)