2019.02.23-携程-Apache Kylin 在携程大数据平台的优化及演进

1 .Kylin —— 

2 .

3 .ART Adhoc l Hive lShark l Hive Shark HDFS 

4 . HBASE Apache Kylin SQL 

5 . 35 769 b l O / T l +<2 a K A l+<2 e %0 l ( C l +<2 L C l c l P) 8 ( P) 8 

6 . D OPS Artnova Adhoc DPP / Hickwall SQL Kylin Clickhouse Presto SparkSQL/Hive Zeus HBASE HDFS 

7 .23 .1 

8 .1 2 . 3 3 

9 .l M l K ( ) l H l ) H l 

10 .l IL K cb l ) .8- g 3 3 l N HA YMa e l 2 KYLIN-3810 l N 1- L ( 

11 .1 f 0-2 35 , 5 -32 3 2 ef 0-2 , 0- 2 22 5 - 3 4 5-3 0-2 , 0- 2 5 5- 2 5 , m h 0-2 .3 5 5 i Kf pd 0-2m 0-2 .3 ,- 35 -2 5 - 0 i nbk ,- 35 0 2 4 g a c , 23 5 -32 5 5 33 4 5 j ol 

12 . . ) ( T051 2 8 T051 2 8B 4. +3.2 51 2 K K 6 D D K SH K SH O 6 O : 

13 . . () - Streaming cube lCube kafka partition offset Kylin Web UI Hermes lBuild-memory Kylin Hbase l batch hive Kafka l NRT MR/SPA RK 

14 . . - l m j l na l e M h l i c g l b o l R 

15 . ( ) . 1 : 5 

16 . K , K , K : , K 

17 .wei_za@ctrip.com