What might cause linked server query to fail when executed using SqlCommand?

So it appears there were two issues here:

1. that queries executed in SSMS behaved differently to those in .NET via SqlCommand, and
2. that MSDTC requires more knowledge than just where to tick the box

SSMS vs .NET

When I was testing via SSMS I was trying to replicate the .NET approach as best possible.

.NET code

var queryText = @"INSERT INTO RemoteServer.RemoteDB.dbo.tst_Address (Address, FirstName, LastName, DateOfBirth)
    SELECT t1.Address, t1.FirstName, t1.LastName, t1.DateOfBirth  
    FROM LocalDB.dbo.tst_Address t1 
    LEFT JOIN RemoteServer.RemoteDB.dbo.tst_Address t2 ON t2.Address = t1.Address
    WHERE t2.Address IS NULL"
using (var trans = cxn.BeginTransaction())
{
    try
    {
        var cmd = this.BuildSqlCommand(queryText, parameters);
        cmd.ExecuteNonQuery();
        trans.Commit();
    }
    catch (Exception ex)
    {
        logger.Error(ex);
        throw;
    }
}

SSMS query

begin tran
begin try

INSERT INTO RemoteServer.RemoteDB.dbo.tst_Address (Address, FirstName, LastName, DateOfBirth)
    SELECT t1.Address, t1.FirstName, t1.LastName, t1.DateOfBirth  
    FROM LocalDB.dbo.tst_Address t1 
    LEFT JOIN RemoteServer.RemoteDB.dbo.tst_Address t2 ON t2.Address = t1.Address
    WHERE t2.Address IS NULL
commit

end try
begin catch

    SELECT  
        ERROR_NUMBER() AS ErrorNumber  
        ,ERROR_SEVERITY() AS ErrorSeverity  
        ,ERROR_STATE() AS ErrorState  
        ,ERROR_PROCEDURE() AS ErrorProcedure  
        ,ERROR_LINE() AS ErrorLine  
        ,ERROR_MESSAGE() AS ErrorMessage;  
    if @@TRANCOUNT > 0
        rollback

end catch

The catch with SSMS is that by default the query is not required to be distributed and doesn't involve the DTC unless it is explicitly told to by using begin distributed tran. Once I added the distributed keyword then SSMS also kindly failed with similar errors to .NET.

DTC configuration

The well enough documented tick boxes are found in Component Services (dcomcnfg from command line) and the typical settings are shown below.

What caught me out was that the Windows firewall rules, while preexisting, are disabled by default. Potentially the difference between the two computers that were attempting to access the RemoteServer was that one had the firewall disabled so was not blocked by the outbound rule, but both were blocked by the inbound rule disabled on the RemoteServer. Here are the outbound rules

and inbound rules that are shown now enabled.