If I understand correctly, you have a flow that goes from SQL Server -> S3. In the output dataset, the 'date' fields are shown as date fields as expected, but in the input dataset they appear as integer columns? Can you clarify what format your input date fields are in?
And have you tried checking the following options in the Settings of the input dataset?
If that doesn't help, can you attach a screenshot of the issue with the MS SQL Server dataset?