Ask HN: Have you used big data ecosystem or MPP DBs to building DWH?

[popfalushi]

Have anyone used big data ecosystem (HDFS, SparkSQL/Impala/Kudu/Drill/etc, HBase/Cassandra/etc) in creating DWH? What are your thoughts?

Nowadays there is a great hype about big data products, BI tools started to gain support for connectivity with SparkSQL, Hive. Spark summit had several people who told tales about their usage of SparkSQL for DWH.

I suppose different stacks have their own usecases. I am inclined to believe that big data stack can be very suitable for processing events, documents with little usage of joins, RDBMS stack (+ HOLAP/MOLAP) are better for DWH for relational data, where there are many interconnected entities and everything is structured. However there is also Greenplum (Postgresql 8.3 fork with transparent sharding support and distributed queries) which is smth in between I guess.

So, if you have experience of building DWH with non-traditional approach, please describe your experience and usecase in comments - I am sure there are many people interested in what you can say.