SQL Server now supports the vast majority of data types in the In-Memory OLTP engine. In fact, the list of supported data types is now so long that it is easier to list the unsupported types instead. The data types that are currently unsupported are:
- Datetimeoffset
- Geography
- Geometry
- Hierarchyid
- Rowversion
- Xml
- Sql_variant
This means that (except for XML), Large Binary Objects (LOBs) are supported, and this covers the max data types varchar(max), nvarchar(max), and varbinary(max). This is great news, as there are many cases where disk-based tables that had only one LOB column, but multiple supported columns, were barred from the In-Memory OLTP engine.
LOB data type support is not limited to tables and indexes—it is now also possible to use LOB data types in memory-optimized stored procedures too. This will allow parameters to be passed into and out of natively compiled stored procedures.
Interestingly, as of SQL Server 2017, the JSON data type is supported. This is possible because JSON is not a native data type like XML, but is stored as nvarchar data, which is listed as a supported data type. This is yet another reason to accept JSON as a possible unstructured data storage type over XML in future SQL Server programming.
An exhaustive list of current In-Memory OLTP limitations can be found on Microsoft Books Online: https://msdn.microsoft.com/en-us/library/dn246937.aspx