Caching

AVEVA PI System Connector 3 (PSC3) includes a built-in caching mechanism that protects data integrity when downstream systems are temporarily unavailable.

When the connector cannot deliver data to the broker or destination, such as broker outages or during network interruptions, data is automatically buffered instead of being lost.

Caching is handled primarily at the connector and operates transparently. It preserves data order and ensures consistency for both PI Data Archive data and Asset Framework (AF) objects. When connectivity is restored, stored data is automatically sent again, and regular data transmission continues without any need for user action.

Caching occurs under the following conditions:

• Caching interval is exceeded

  When operating in real-time mode, PSC3 switches to disk caching mode if data remains in memory for 30 seconds without transmission.

  In History Recovery mode, disk caching mode is activated after data has been stored in memory for 90 seconds without being sent.

  The real-time (RT) and history recovery (HR) queues are monitored independently.

• Status of Broker is busy

  If the next broker in the chain receives messages more quickly than it can handle, leading to the unread messages exceeding a specific limit, the preceding component switches to disk caching mode.

  When the broker returns a busy status, the connector reduces its activity and might start caching data to disk.

  See Broker health and flow control for additional information.

• Retry limit reached

  If data transmission to the broker fails three times consecutively, the connector switches to disk caching mode, even if the connection is established.

This section describes where cached data is stored, how recovery and replay are handled, and key operational considerations for managing cached data.