Future of Data virtualization s/w in Big data world

There are data virtualization s/w such as Cisco composite, Denodo, Informatica data service are being used by enterprise. They all promised to connect to heterogeneous sources to brings enterprise reporting under a uniform business layer(logical). Although I have not seen these tool fulfill their promises fully(heterogeneous joins are very slow), I am more skeptical about future of these tool in the world of big data.

Could you please share your thoughts on future of these tools and what may be best open source tool for data virtualization in big data world?