Got your point. The fact is that, once the cube segment has been built, kylin itself will NOT do any clean up things.(the only one clean-up job is to delete the intermediate data) Currently kylin does not support partially clean-up the segment data, if users find the cube obsolete, you can disable and purge the cube which will drop the hbase table
Best Regard Zhou QianHao On 4/23/15, 5:28 PM, "[email protected]" <[email protected]> wrote: >Hi, QianHao > >Agreed for your explaination of hierarchical dimensions and validness >check. What I am concerned is : > >Does kylin do some work for settings of the cube segment table TTL? As a >common sense, we would like to > >clean up the day cube segment data after 30 days, right ? Or kylin did >not do this work and we need to delete > >the data manually ? > >Thanks for your prompt reply. > >Sun. > > > >[email protected] > >From: Zhou, Qianhao >Date: 2015-04-23 17:24 >To: [email protected] >Subject: Re: Kylin hbase table for data deletion >1. Kylin create one hbase table for one cube segment(one cube may have >several cube segments, depend on incremental build and merge) >2. For hierarchy levels like day/month/year, if you want to query base on >day, you will first have to explicitly declare month and year(otherwise >this day will be meaningless). Kylin also store the data the same way > >Best Regard >Zhou QianHao > > > > >On 4/23/15, 4:02 PM, "[email protected]" <[email protected]> wrote: > >>Hi , all >> >>Note that Kylin would only create one hbase table for the cube. If we had >>hierarchy levels of dimensions as day, month, year, >> >>the day cube should be set a TTL as 30 days sort of , right ? Thus would >>cause a self deletion after 30 days and the other cube >> >>data should also be considered as setting TTL. >> >>Question is : how kylin deal with such condition ? If not, could cause >>severe pressure with a long term run. >> >>Best regards, >>Sun. >> >> >> >>[email protected] >
