CONNECT to PI Agent

CONNECT to PI enables you to transfer data from CONNECT data services SDS streams to an on-premises Data Archive. This can be used for many applications, such as performing analysis in the cloud and sending the results to a PI Server.

CONNECT to PI architecture

The diagram below shows the flow of data when using CONNECT to PI.

The CONNECT to PI Agent communicates with CONNECT data services via REST calls and with the PI Server via the OMF endpoint on the PI Web API.

The agent reads SDS stream data from CONNECT data services to PI and writes OMF stream data to PI Points.

The agent sends health and diagnostics data to CONNECT data services which is presented in the CONNECT to PI Agent portal and is available as SDS stream data.

The agent does not have to be in the same network as the PI Server, but it does need to have access to PI Web API and internet access to CONNECT data services.

Restrictions of the CONNECT to PI architecture

These are a few restrictions with the existing CONNECT to PI architecture:

• Only one CONNECT to PI Agent can be installed on a given host computer.

• The CONNECT to PI Agent transfers stream data from a namespace to PI Points in a Data Archive. No data is transferred to a PI AF Server.

• Each CONNECT to PI Agent is tied to one Data Archive through the specified PI Web API endpoint.

• The CONNECT data services namespace you select during the agent configuration process is used to store the agent and transfer configuration. However, you can configure a transfer to use any namespace within the given tenant.

• A transfer sends data from a single namespace to a single Data Archive. Multiple transfers can be configured from multiple namespaces within the given tenant, using a single agent, but each transfer sends data to a single Data Archive. A second agent on a second host computer can be configured to transfer data from one or more namespaces to a different Data Archive, if required.

CONNECT to PI best practices

• We recommend configuring buffering between PI Web API and the Data Archive. For Data Archive collectives, buffering is required. Ensure the PI Buffer Subsystem is configured on the PI Web API node following the procedure detailed in Buffering services.

• If PI Point compression is typically used in the Data Archive, it should be configured after the PI Points are created by the agent. Compression can be set in bulk using tools such as PI Builder. See Edit PI Points. However, no compression will occur, regardless of the PI Point configuration, if the PI Buffer Subsystem is configured. Thus, a decision on the importance of buffering versus compression is required.

• If you are transferring large amounts of historical data from CONNECT data services, we recommend you increase the buffer size of the agent to avoid buffer overrun. The buffer size is adjustable using the maxBufferSizeMB property in the agent configuration. Increase the buffer size to maximize the buffering capacity while not exceeding the amount of available disk space on the computer. See CONNECT to PI Agent buffering.

• For heavy workloads, install the CONNECT to PI Agent on a host computer that is separate from the PI Web API and Data Archive host computers.

• Keep the CONNECT to PI software version up-to-date. The portal indicates when an agent is out of date and needs to be updated.

• CONNECT to PI should not write to the same PI tags that PI to CONNECT reads from. As an example, if you are transferring Tag A from a PI Server to CONNECT data services with PI to CONNECT, you should not configure CONNECT to PI to write data from CONNECT data services back down to Tag A.