ODBC -vs- Replication

Hi --

I have a client interested in accessing Progress data from a SQL app. What questions should I be asking in order to determine if it would be better to set up ODBC access so that the Progress db appears as a linked server in SQL, versus setting up dataserver/replication to populate SQL tables from Progress either periodically or real-time?

Pros/cons I thought of:

• If we use ODBC, we will want to set up database security, which will need to be administered; replication is outbound-only so no security impact.
• Under ODBC, there will be network traffic; if several users query the same data, the same bytes will traverse the network repeatedly; whereas with replication no byte goes over the wire twice, i.e. once data is in SQL, it stays local.
• ODBC is a one-time setup and the work of designing "good" queries falls to non-Progress folks. Replication will take significant setup, but the SQL copy can have its own structure, indexes, field names, etc...

Your thoughts?

This is 11.6 on RHEL7.