I've had my doubts about Aster, but you have to give Teradata credit
I've gone from excited to disillusioned and back to somewhat interested in Teradata's strategy for Aster. When Teradata first acquired in 2011, I thought it was a good move that would broaden Teradata's capability to MapReduce-based analysis including time-series, text, and other approaches all expressed in SQL. When Hadoop really took off, I started thinking, why would anybody want to run Teradata and Aster and Hadoop? That's three MPP environments, and it's not trivial or inexpensive running any one of them.
Given the latest advancements in Aster -- adding graph and R-based analyses to the MapReduce and SQL options already available -- I now see Aster as an option for Teradata shops that want all those analysis options sooner, rather than later. The key draw is that all of these analysis approaches are handled in SQL. So Aster users can count on SQL expertise instead of hiring the types of experts that would be needed to master each of the engines available within, say, Apache Spark. That's the billing, anyway. Anybody have insight on the cost of Aster vs. running Spark or the ease of mastering all the analysis options in Aster vs. those available on Spark? The devil is in the details.