The Portfolio datatypes in the Financial Toolbox cannot use Global Optimization solvers. On the other hand you can use a Global Optimizer which in turn calls the Portfolio optimizations. I have seen this used in several different scenarios.
The benefits of using the Portfolio datatypes over CVX is that it is using a convex solver already and setting your problem up is easy. CVX is just trying to make the MATLAB code for general convex problems easier, but does not contain pre-defined types of constraints (group constraints, tracking error, etc.). Furthermore CVX is not capable of solving using non-derivative based solvers. So for instance if you have a non-smooth problem you may wish to go the route of Global optimization.