re: EMC Brings Data Analysis Breakthrough To Hadoop
The cost savings discussions are quite often around the per-terabyte cost of Hadoop vs. more conventional relational platforms. That's not really the issue here. It's assumed that Hadoop is the scale champion and that it's capable of handling multistructured data -- something relational databases don't do well.
With this announcement EMC is addressing the same Hadoop flaw that Cloudera is addressing with Impala -- supporting SQL querying that is faster and broader than the primitive, SQL-like querying supported by Hive. Yes, speed of query is important, but just getting access to standard-SQL querying is a big win. This is something for people who are already sold on Hadoop... not for those who are still tire kicking and wondering if they should even get into big data.