Agreed, with the current processing frameworks like Apache Hadoop and Spark, a Nation wide network of commodity hardware machines, we could create an efficient yet very powerful cluster grid that would widely available for all sorts of research work. That's cheap and effective, reminding me of...