The "routing to Hive" feature is deprecated as its user experience is bad (user doesn't aware it was routed, would think Kylin is hung).
The alternative way is building such a router atop Kylin, which should be simple like a try-catch block; 2016-08-29 23:54 GMT+08:00 Sarnath K <[email protected]>: > It looks like this has to be tested through only a Java program using JDBC > driver. The SQL interface in Webapp, does not seem to support cube misses > and backend hive handling. > We will try again tomorrow. > > Today, I also verified the hive.URL parameter in the source in > KylinConfigBase.java. but it has a note that the methods returning the hive > URL, user and passwd are deprecated. What is the alternative planned in > their place? > Thanks, > Best, > Sarnath > On Aug 28, 2016 8:36 PM, "Sarnath K" <[email protected]> wrote: > > > This is cool. I will have this tried this week. Thanks! > > On Aug 28, 2016 7:36 PM, "ShaoFeng Shi" <[email protected]> wrote: > > > >> The current implementation only route to hive, via Hive JDBC (add > >> hive.url, > >> hive.user, hive.password in kylin.properties); but it should be doable > to > >> extend; Please open a JIRA to track this requirement, thanks! > >> > >> 2016-08-28 15:40 GMT+08:00 Sarnath K <[email protected]>: > >> > >> > Yes, I am looking at routing cube-misses to impala. This can > >> significantly > >> > increase net performance. Impala can execute queries on Hive tables as > >> > well. Impala provides a JDBC driver as well. > >> > If this is possible, I just want to know what config variables I need > to > >> > set to make this happen. > >> > Thanks for your time! > >> > > >> > Best, > >> > Sarnath > >> > On Aug 28, 2016 7:10 AM, "ShaoFeng Shi" <[email protected]> > wrote: > >> > > >> > > I guess Sarnath's data source is still in hive, he wants to route > some > >> > > queries (cube didn't catch) to Impala, instead of hive. @Sarnath, > can > >> you > >> > > confirm? Thanks. > >> > > > >> > > 2016-08-28 7:20 GMT+08:00 Li Yang <[email protected]>: > >> > > > >> > > > Use Impala as input source requires code change. It's not just > some > >> > > > configuration. Kylin is designed to support multiple data sources > >> > however > >> > > > implementation must be provided for each kind of data source. > >> > > > > >> > > > On Fri, Aug 26, 2016 at 8:35 PM, Sarnath K <[email protected]> > >> wrote: > >> > > > > >> > > > > Thanks for your answer and time. I would be happy to give a try. > >> But > >> > > what > >> > > > > config value I need to change (the above one is okay?) And from > >> which > >> > > > > version of Kylin is that config value supported? Sorry for being > >> > > ignorant > >> > > > > here. Any help is appreciated. i have the infra with me. Just > >> tell me > >> > > > what > >> > > > > config value I should set in kylin.properties? > >> > > > > On Aug 26, 2016 1:44 PM, "ShaoFeng Shi" <[email protected] > > > >> > > wrote: > >> > > > > > >> > > > > > Hi Sarnath, > >> > > > > > > >> > > > > > Routing to Impala looks better than routing to Hive; In theory > >> it > >> > is > >> > > > > doable > >> > > > > > as Kylin use standard JDBC API, but we haven't tried that; you > >> can > >> > > > have a > >> > > > > > try and share your finding with the community, thanks! > >> > > > > > > >> > > > > > 2016-08-25 16:05 GMT+08:00 Sarnath K <[email protected]>: > >> > > > > > > >> > > > > > > Or in other words, > >> > > > > > > Can I specify Impala JdBC server in kylin.Hive.jdbc.driver > ??? > >> > > > > > > There is a comment saying 'not in use' in the Github > >> > > kylin.properties > >> > > > > > file > >> > > > > > > under examples.... > >> > > > > > > > >> > > > > > > Any inputs would be great! > >> > > > > > > > >> > > > > > > Best, > >> > > > > > > Sarnath > >> > > > > > > On Aug 25, 2016 11:30 AM, "Sarnath K" <[email protected]> > >> > wrote: > >> > > > > > > > >> > > > > > > > Hi, > >> > > > > > > > I tried searching for this a lot but could find a > convincing > >> > > > answer. > >> > > > > > > > > >> > > > > > > > Can Kylin use Impala JDBC driver underneath to run queries > >> on > >> > > Hive > >> > > > > > table > >> > > > > > > > instead of Hive server? > >> > > > > > > > > >> > > > > > > > If this is not possible, I would like to understand why > >> this is > >> > > not > >> > > > > > > > possible and if there is a roadmap for this. > >> > > > > > > > > >> > > > > > > > Thanks, > >> > > > > > > > Best, > >> > > > > > > > Sarnath > >> > > > > > > > > >> > > > > > > > >> > > > > > > >> > > > > > > >> > > > > > > >> > > > > > -- > >> > > > > > Best regards, > >> > > > > > > >> > > > > > Shaofeng Shi > >> > > > > > > >> > > > > > >> > > > > >> > > > >> > > > >> > > > >> > > -- > >> > > Best regards, > >> > > > >> > > Shaofeng Shi > >> > > > >> > > >> > >> > >> > >> -- > >> Best regards, > >> > >> Shaofeng Shi > >> > > > -- Best regards, Shaofeng Shi
