Thanks very much, you means I have to create a temporary table in hive and use this table to create cube in kylin, what's more, I can not drop this table because when I want to build incremental cube I need select those records from source table and insert into the temporary table ..OR I can create new table whose PK is timestamp and I take this table as a dimension table. I think this would work fine.
However, I actually hope kylin can support something like keyExpression(you can define dimension level with any column expression ) in mondrian when create a dimension level rather than only can specify a column.I think time is a special dimension and it will more easy to use with this feature.. 2015-09-01 17:17 GMT+08:00 Xiaoyu Wang <[email protected]>: > Hi Feng! > You can transform the timestamp column to day,hour two columns in hive > before you create cube. > Then in cube design. You can add hierarchy dimensions like day,hour. > In the sql: select day,hour,sum(sales) from table group by day,hour > > > 在 2015年09月01日 17:07, yu feng 写道: > >> Hi all ： >> I have use kylin for about one month，but can not find how to generate >> time hierarchy based on my timestamp column， I have an HIVE table like >> this: >> >> CREATE TABLE `sales_table`( >> `time` timestamp, >> `sales` int) >> >> and I have a timestamp column named time, I want to get result like this : >> >> select date(time) ,hour(time),sum(sales) from sales_table group by >> date(time), hour(time); >> >> 2015-08-27 16 18388 >> 2015-08-27 17 35301 >> 2015-08-27 18 41884 >> 2015-08-27 19 40638 >> 2015-08-27 20 40294 >> 2015-08-27 21 43753 >> >> i.e. you can generate full time hierachy like year/month/day/hour.. when >> you have a timestamp column. >> >> I have no idea how to achieve it with kylin and I will appreciate it very >> much if someone can help me，Thanks~~ >> >>
