Thank you Alberto! Direct implementation on HDFS will be most ideal, removing all the unnecessary dependencies. However that also means Kylin will be on its own to do more storage level tasks like indexing and fast random lookup etc..
A very good option to look into. Cheers Yang On Sun, Mar 19, 2017 at 6:26 AM, Alberto Ramón <[email protected]> wrote: > Hi > > I saw the PPT of Apahe Kylin 2.0 for the Haodoop Summit of Li Yang > > In the "What is the next" section, can be interesting for Apache Kylin, > evaluate Apache Ozone (PPT of Summit 2015 > <http://www.slideshare.net/Hadoop_Summit/ozone-an-object-store-in-hdfs> > Video <https://www.youtube.com/watch?v=jnjSb9GEnO4> HDFS-7240 > <https://issues.apache.org/jira/browse/HDFS-7240> ) same ideas than Kudu > but implemented in HDFS directly and perhaps an "HBase Killer" for some > uses case > > BR, Alb >
