PI interfaces and high availability

Interfaces are the components of the PI System that collect time-series data from data sources and send the data to Data Archive for storage. To implement HA you will need to configure interfaces to support failover and n-way buffering. Failover ensures that time-series data reaches Data Archive even if one interface fails; n-way buffering ensures that identical time-series data reaches each Data Archive server in a collective.

To support failover, install a redundant copy of an interface on a separate computer. When one interface is unavailable, the redundant interface automatically starts collecting, buffering, and sending data to Data Archive. To support n-way buffering, configure the buffering service on interface computers to queue data independently to each Data Archive server in a collective.

In some deployments, interfaces send output points (that is, data from Data Archive) to the data source. With proper configuration, failover considers the availability of Data Archive for output points in addition to the availability of the interface.

Buffering for PI interfaces also has different implications in high availability scenarios. We recommend n-way buffering for interfaces to protect against data loss. For a description of n-way buffering and implementation instructions, see N-way buffering for PI interfaces.