Hi Andras, Is this specific to real-time streaming, or it is also related to the normal batch cube? I want to narrow down the scope. Besides, how complex the queries are? I think that may need more inputs. If you can bring some sample queries that would help to some extend.
Best regards, Shaofeng Shi 史少锋 Apache Kylin PMC Email: [email protected] Apache Kylin FAQ: https://kylin.apache.org/docs/gettingstarted/faq.html Join Kylin user mail group: [email protected] Join Kylin dev mail group: [email protected] Andras Nagy <[email protected]> 于2019年11月28日周四 下午9:06写道： > Dear All, > > We are troubleshooting slow queries in our Kylin deployment, and we > suspect that the issue is not with the cube definitions, but with our > queries. At least we have some quite complex queries with a lot of range > checks on dimension values, and we have observed different response times > by changing the queries to alternative, but functionally equivalent ones. > > Although it's hard to come to conclusions because we see a large variance > in query response times (for the same query in the same environment, at > roughly the same time). > We have disabled query caching in kylin.properties > (kylin.query.cache-enabled=false) to be able to have more conclusive > results on what effect certain changes have on query execution time, but we > still observe variance in query results on an environment that otherwise > has no load. Perhaps this is due to caching within HBase or within the > streaming receiver. > > Do you have any guidelines, best practices, documentation on how to tune > queries for Kylin? (I'm aware of some cube tuning guidelines from the Kylin > documentation, but now I'm looking for advice specifically about query > optimization.) > > Many thanks, > Andras >
