Hi Luke, Thanks. Mondrian is an OLAP server by itself (as I read from wiki). So I don't understand what connecting Mondrian to kylin actually means. In fact, if we can connect Mondrian to hiveserver then we can get the aggregations done and served through MDX but I am not sure whether it will be scalable. I also discovered the olap4j spec and an MDX parser available ( http://www.olap4j.org/api/org/olap4j/mdx/parser/MdxParser.html) Calcite is all about SQL. But it is good to know this. Thanks for the info. Best, Sarnath On May 19, 2015 5:07 PM, "Luke Han" <[email protected]> wrote:
> Hi Sarnath, > You are right, sometimes MDX is more easy to use for some scenarios > especially for complex analytics. > Using Mondrian is one idea, but as Jakob said, there's should be some > issues have to be fixed to make it work. > And, expose MDX interface is actually doable thing from Kylin, based on > cube's metadata (dimensions, measures and all information already there). > The issue is you have to figure out a way to parse MDX query and call > underlying storage. You could check Calcite first to see if MDX has been > supported or not, it will be good idea to leverage Calcite for this also. > > Thanks. > > Luke > > > Best Regards! > --------------------- > > Luke Han > > 2015-05-19 13:12 GMT+08:00 Sarnath <[email protected]>: > > > Hmm..Interesting. From what I browsed, I thought Mondrian is a DMX based > > OLAP tool. Since kylin is SQL based, how did you get it working (at > least. > > Unstably)? Thanks for your time. > > On May 19, 2015 12:52 AM, "Jakob Stengård" <[email protected]> > > wrote: > > > > > You could use Mondrian with kylin as backend, maybe. > > > > > > I tried it a few months ago. Didn't work so well then, but maybe kylin > is > > > more stable now. > > > > > > Best Regards > > > Jakob Stengard > > > > > > On Mon, May 18, 2015 at 7:57 PM, Sarnath <[email protected]> wrote: > > > > > > > Yes, I agree. SQL is more commonplace than MDX. > > > > > > > > However, Sometimes MDX is hidden under an OLAP tool... The end-user > may > > > > never know it. And having MDX support would make sure Kylin connects > to > > > > several existing BI tools out of the box... > > > > > > > > Actually, I was doing a simple aggregator project without realizing I > > was > > > > into OLAP. And then I started reading up and found Kylin - which kind > > of > > > > already does what I was thinking to do...and much more... So, I was > > just > > > > wondering if I could serve up my aggregations via MDX/ODBO/XML for > > > > Analysis. I believe that would help me to connect the aggregations to > > > OLAP > > > > tools right away... Do you have any recommendations on how this can > be > > > > done? > > > > > > > > Best, > > > > Sarnath > > > > > > > > On Mon, May 18, 2015 at 11:15 PM, Han, Luke <[email protected]> wrote: > > > > > > > > > The design thinking behind of this is there could be more cubes to > > > serve > > > > > different SQLs with same star schema, and only expose Table and SQL > > to > > > > user > > > > > is more easy way for people to onboard and learn and integrated > with > > > any > > > > > SQL based tools. > > > > > > > > > > And, how many usage of MDX compare to SQL? And how many people know > > SQL > > > > > but not MDX:-) > > > > > > > > > > Thanks. > > > > > > > > > > > > > > > > > > > > > > > > > 发自我的 iPhone > > > > > > > > > > > 在 2015年5月19日，01:32，Sarnath <[email protected]> 写道： > > > > > > > > > > > > Hi, > > > > > > > > > > > > I am very new to Kylin. > > > > > > > > > > > > From the architecture of Kylin - what I understand is that - > Kylin > > > does > > > > > all > > > > > > the aggregations but does not expose the result as > > multi-dimensional > > > > > > database. Instead, it exposes the underlying as a normal > relational > > > > > > database - but optimizes queries by serving from the aggregations > > > > > wherever > > > > > > possible.... > > > > > > > > > > > > Is there a reason why Kylin chose not to go with MDX or ODBO...? > > Did > > > it > > > > > > choose ODBC only because of Tableau? > > > > > > > > > > > > Appreciate any clarity on this, > > > > > > Thanks, > > > > > > Best, > > > > > > Sarnath > > > > > > > > > > > > > > > > > > > > > -- > > > > > > Med vänlig hälsning (Best Regards) > > > *Jakob Stengård* > > > > > >
