There are some good reasons why you may want to explore using heterogeneous replication, but the best reason is that it allows an administrator to offer SQL Server to clients whose companies have a critical investment in other data systems.
SQL Server lets you replicate data from any source that you can link to with a 32-bit ODBC driver. For Windows 2000, NT and even Windows 98 you can use an OLE DB driver instead. Thus you can use SQL Server to replicate data sources as diverse as DB2 AS400, DB2/MVS, Oracle, Exchange and Access. You can also replicate data between different versions of SQL Server and even with SQL Server for Windows CE.
To publish heterogeneous data using either ODBC or OLE DB you would apply a push subscription to the subscriber from the publisher. Subscribers can be any of the aforementioned data sources. You can also have SQL Server subscribe to replicated data from these heterogeneous sources.
Replication can be published as a snapshot or as a set of transactional records. You might want to use this scheme when you are using SQL Server as an analysis tool in data warehouses. SQL Server can collect all of the records by replication and then either send the data to a consolidated data warehouse or mart, or use the data as a multidimensional database with the SQL Server Analysis Services.
If you implement a heterogeneous replication scheme you will need to program the Replication Distributor Interface using SQL-DMO or use a third party application for that purpose. To review how to program replication using heterogeneous data sources and various programming options you may want to consult Microsoft's MSDN explanation here.
Barrie Sosinsky is president of consulting company Sosinsky and Associates (Medfield, Mass.). He has written extensively on a variety of computer topics. His company specializes in custom software (database and Web related), training and technical documentation.