Read the official website and write where you learn
1, the need to first establish model
2, Kylin need to configure the fact table, latitude table; You can customize the join. My usage differs from the official advice, I am directly in hive to take all the join into a single table, and then according to a single table Cude Kylin. Because my join has some business processing at the same time.
3. You need to select the Latitude field of the Cude and the Measure field of the aggregation; the Latitude field can be selected in all tables, and the Measure field can only be selected in the fact table (the field of measure is Sum,avg,count)
4, the establishment of model, you need to choose partition, is generally incremental by the day.
5, Cude, can be used according to the needs of the use of "hierarchy" and "derivation" to optimize
1), level, add dependencies between fields, can only combine fields to cube, reduce the complexity of cude
2), derive, combine multiple fields into a single field (that is, the primary key) and query according to the primary key.
3), combination, specify a combination of fields, cube by combination of cube, reduce the complexity of cube
6.
Kylin Study Notes