http://code.google.com/p/olap4cloud/wiki/UserGuide mentions their comparison to Hive... I wonder just how tuned their Hive is to take so much longer than this layer, even without the pre-aggregation. Would row order storage make that much of a difference? Doesn't Hive now have some type of indexing?
Afaik hive has no indexing at all. That's part of why it's so simple (well, for hadoop folks) to use. All you need to do is copy a bunch of csv, tsv, logs, whatever in a HDFS path, tell hive some basic infos about the files and you can join the table with any other information in the system. Really powerful stuff.
However, if you don't have any legacy data and/or don't mind manual import into HBase then something like the above solution may reduce the complexity of the issue enough to get much better query performance.
