Issuing Inserts/Updates to Progress OpenEdge via ODBC

I'm attempting to build some ETL to integrate our SQL Server environment with a remotely hosted 3rd party app running Progress OpenEdge. I have a linked server to it set up in SQL Server 2014. Selects are working well enough, though I find that using openquery is much, much faster than the simpler select * from SERVER.DB.SCHEMA.TABLE syntax. I'm struggling with writes, however. I've found several references for valid update syntax for linked servers, but these all give permissions errors: select * from openquery(SERVER,'update FOO.BAR set "COL" = 0 where "COL" = 1') update fb set [COL] = 0 from SERVER.DB.FOO.BAR fb where [COL] = 1 update openquery(SERVER,'select * from FOO.BAR where "COL" = 1') set [COL] = 0 This syntax is the only one I've found that works: declare @script nvarchar(max) = N'update FOO.BAR set "COL" = 0 where "COL" = 1' execute (@script) at SERVER This works for both inserts and updates, but I can't see a way to load a set of data this way, because I can't reference my local tables. In SSIS, an ODBC connection passes the connectivity test, but fails during initialization: > Error: 0xC0014020 at Package1, Connection manager "SERVER": An ODBC error -1 has occurred. > > Error: 0xC0014009 at Package1, Connection manager "SERVER": There was an error trying to establish an Open > Database Connectivity (ODBC) connection with the database server. > > Error: 0x20F at Data Flow Task, ODBC Destination : The > AcquireConnection method call to the connection manager SERVER failed with error code 0xC0014009. There may be > error messages posted before this with more information on why the > AcquireConnection method call failed. > > Error: 0xC0047017 at Data Flow Task, SSIS.Pipeline: ODBC Destination > failed validation and returned error code 0x80004005. > > Error: 0xC004700C at Data Flow Task, SSIS.Pipeline: One or more > component failed validation. > > Error: 0xC0024107 at Data Flow Task: There were errors during task > validation. Ultimately, I'm searching for a better way to merge/upsert a set of data (anything from a few dozen rows to a couple hundred thousand) without resorting to using RPC to insert/update one row at a time.