Custom virtual table data providers
Unsure about entity vs. table? See Developers: Understand terminology in Microsoft Dataverse.
Using the Microsoft Dataverse Data SDK, .NET Developers have the option of creating custom virtual table data providers to help integrate external data source types that are not supported by an existing data provider. Each data provider is composed of a reusable set of Dataverse plug-ins that implement the supported CRUD operations. For each virtual table, also known as a virtual entity, developers can create plug-ins and register them representing each of the Create, Update, Retrieve, RetrieveMultiple and Delete operation. This section provides fundamental information about data providers and approaches to developing custom providers, including example code.
As an alternative to creating a custom data source provider, you should consider adapting your data source to an existing data provider. For example, if you create an OData v4 interface to your external data source, then you can directly access it with the supplied standard OData v4 Data Provider, which supports CRUD operations as well. The mechanism of adding this REST interface varies with the underlying data service technology, for example see WCF Data Services 4.5. OData has broad industry support, with a wide range of dedicated tools and compatible technologies.
Custom data providers require substantial development resources to create and maintain. You must have fundamental knowledge of the following areas:
- The external data source schema and associated data access techniques. This domain knowledge is specific to the external data source type.
- Dataverse definition schema: More information: Work with table and column definitions using code.
- Dataverse event framework: More information: Event Framework.
- Dataverse plug-in architecture and development: More information: Use plug-ins to extend business processes.
Microsoft.Xrm.Sdk.Data.dll assembly is available as a NuGet package: Microsoft.CrmSdk.Data
Categories of providers
There are two general categories of data provider you can create using the virtual table data SDK assemblies: generic or targeted. The table below describes these approaches and matches them to the data provider development model best suited for that approach.
|Generic||"Bare metal" provider||These providers can flexibly translate FetchXML query expressions to the associated request to the external data source, then return the resulting table instances. Such a provider can be reused for all instances of this data source type. This approach is the most general but is more complicated to develop. If the schema of the data source changes, the affected virtual tables must only be remapped.|
|Targeted||LINQ provider for known schema||Such a provider only narrowly translates queries into the associated LINQ call to a known, existing data source instance. The data source must be a LINQ provider as described in the topic Enabling a Data Source for LINQ Querying. This approach is limited to a specific data source instance, but requires much less coding. If the schema of the data source changes, the data provider must be updated and rebuilt.|
The standard OData v4 Data Provider and the Cosmos DB Data Provider are examples of generic providers.
Steps to use a custom data provider
There are several steps that are required to create a virtual table data provider solution that can be imported into your Dataverse applications:
- Develop the custom data provider plug-in DLL (or set of DLLs).
- Register the custom data provider with your Dataverse service using the Plug-in Registration Tool (PRT).
- Create a data provider solution.
- Customize the data source table to reflect your data type or specific instance.
- Export the custom data provider solution.
Because virtual tables support CRUD operations, you will write the data provider in the form of a plug-in registered on the Create, Update, Retrieve, RetrieveMultiple and Delete events. Each respective event will include information in the execution context which describes the kind of data to return.
|Retrieve||Describes which table to retrieve as well as the columns and any related tables to include.|
|RetrieveMultiple||Contains a QueryExpression object defining the query. The framework contains a QueryExpressionVisitor class designed to inspect different parts of the query expression tree.|
For both events, you must :
- Convert the respective information in the execution context into a query that will work for your external data source.
- Retrieve the data from the external system.
- For Retrieve, convert the data into an Entity; otherwise, for RetrieveMultiple, convert it to an EntityCollection. This result is returned through the Dataverse platform to the user executing the query.
The classes in the Microsoft.Xrm.Sdk.Data namespace provide a framework to assist in mapping the Dataverse query information from the execution context into a query in the format appropriate for your external data source. This framework will help you convert the data returned in to the appropriate Entity or EntityCollection types expected by the Dataverse platform.
Data provider exceptions
If for any reason your code cannot achieve the expected result, you must throw the appropriate error. The Microsoft.Xrm.Sdk.Data.Exceptions namespace contains the following exception classes, derived from SdkExceptionBase, that you can use for this purpose:
|AuthenticationException||An error occurred during security authentication to the external data source service; for example HTTP status 401 received from the external data service. Typically occurs because the current user does not have proper privileges or the connection information in the associated EntityDataSource is incorrect.|
|EndpointException||The endpoint configuration in the data source table is invalid or the endpoint does not exist.|
|GenericDataAccessException||A general data access error, used when the error does not map to a more specific exception.|
|InvalidQueryException||The specified query is invalid; for example it an invalid clause combination or unsupported comparison operator.|
|ObjectNotFoundException||The specified record in the external data source does not exist.|
|TimeoutException||The external operation did not complete within the allowed time; for example, the result of a HTTP status 408 from the external data service.|
Unlike an ordinary plug-in, you will only use the Plug-in Registration Tool (PRT) to register the assembly and the plug-ins for each event. You will not register specific steps. Your plug-in will run in stage 30, the main core transaction stage for the operation that is not available for ordinary plug-in steps. Instead of registering steps, you will configure your data provider using the following tables.
|EntityDataProvider||Defines the plug-ins to use for each event and the logical name of the data source.|
When the definitions for your virtual table is configured, your plug-ins are registered using the PRT and the correct configuration data is set in the EntityDataProvider table, your virtual table will start to respond to requests.
Submit and view feedback for