Stream views

A stream view is a logical overlay that enables you to customize your view of streaming data so it is most useful to you. While you cannot change the properties of types, stream views enable you to create a view of a stream so it appears as if you had changed the type. You create a stream view by choosing a source and target type, and then defining mappings between the properties of the two types. The source type is the type associated with the stream. The target type includes the properties you want to include in the stream view. In effect, you can remove, rename, or add properties without altering the original stream type.

Using a stream view on data retrieval affects only the data retrieved, and does not alter the original source data.

For example, a process engineer and a maintenance technician might want to see different data that exists in the same stream. By creating a stream view, you can change the appearance of the data to meet the needs of both users, without changing the original data. In addition, you can use stream views to convert units of measure and change property names so they are more appropriate for a particular audience.

Use the CONNECT data services portal to set up stream views, or use REST APIs to define stream views programmatically. If you are using the .NET framework, client libraries are available to help create and use stream views.

PI Server counterpart

While there is no direct counterpart for stream views in PI Server, the closest analog is the PI point data reference in PI AF. Like stream views, PI point data references allow you to give a friendly name and unit of measure on top of an existing PI point, effectively viewing it in a different way. PI point data references do not alter the underlying PI point, just as stream views do not change the source stream or type in the Sequential Data Store.

Stream views best practices

The following best practices are recommended when working with stream views:

• Stream views can be useful when you are using PI to CONNECT. PI to CONNECT creates streams with many additional flags, for example IsQuestionable, IsSubstituted, and IsAnnotated. You can use stream views to limit these properties or to give them a more meaningful name.

• When you use a stream view, use caution when not all properties in the target type exist in the source type. Properties that are not mapped return their default value in the target type. For example, if the target type contains a property of data type double that is not mapped, data transformation gives that property a value of 0. Assume you have a source type with the properties, Timestamp and Pressure. You create a stream view that maps to a type with the properties Timestamp, Pressure, and Temperature. If Temperature is of data type double, a value of 0 (zero) is always returned for it, because it cannot be mapped from the source type.

• You can use a stream view to change a stream's underlying type. Use caution if you do so, because any source properties that cannot be mapped to a target property are removed from the stream data. Unmapped target properties are set to their default value for all existing events. For example, if you have a source type that has the properties Timestamp, Pressure, and Temperature, and you use a stream view to map to a type with the properties Timestamp, Pressure, and Depth, you can map only Timestamp and Pressure. In this case, the source data for the Temperature field will be lost, and the existing events will be given a value of 0 for the new Depth property.