We have transactional hourly data and daily bulk loaded data from operational systems which we engineer into data warehouse dimensions (and facts) in a MS SQL Server environment. If recommended, what fast and efficient techniques (Flume, Sqoop, Kafka, HDFS, HBase, Hive, KUDU, Spark, Impla) could we apply to achieve this in Hadoop? The assumption is that these dimensions would be consumed hourly by the MS SQL Server data warehouse.