Let me take these questions out of order, they'll make more sense that way.

>2) What are the decision criteria for using a local/remote view.

If you are using native VFP tables, use local views. If you are using some other data store (eg. SQL Server, SQL Anywhere, Oracle, etc.), use remote views.

>
>1)What are the performance penalties, if any, for using Remote View instead of Local Views.
>

You're adding an ODBC layer on top of your app, so you'll notice the difference in performance. I don't have any numbers to give you, though. There are some cases where remote views for VFP data might make sense, but for most cases there isn't any benefit, and there are penalties.

>
>3) How to create a connection string for the remote view at runtime. Because, the datasource for the clients is going to be different for each computer the software is installed on.

Take a look in the MM's DevGuide (around page 174). There is a section about creating named connections. It's basically the CREATE CONNECTION command.

>
>4)Does the Remote View provide me with a Stateless connection to my data. In other words does the Remote View keep a connection to the data open or hard wired?
>

The framework can/will maintain a single shared connection to the backend.

>5)What is the best option for providing a Row Source from these free tables?
>

Not sure what you mean here.

HTH