+1 Actually, Kylin plays a much more important role than just an OLAP engine in many Kylin users’ production environments.
ShaoFeng Shi <shaofeng...@apache.org> 于2020年1月12日周日 下午8:32写道: > Hello, Kylin developers and users, HAPPY NEW YEAR 2020! > > In last month, we released Kylin 3.0, with the new Real-time streaming > feature and a Lambda architecture. This allows our users to host only one > system for both batch and real-time analytics, and then can query batch and > streaming data together. > > If you look at Kylin's home page, its slogan is still the "OLAP Engine for > Big data", which was made 5 years ago when it was born. While today, > Kylin's capability has been verified beyond an "OLAP engine". I visited > many Kylin users in China, US, Euro in last year, and have got many > different scenarios: > > 1. eBay initiated the Kylin project to offload analytical workloads from > Teradata to Hadoop; Kylin serves the online queries with high performance > and high availability. Till today, Kylin serves millions of queries every > day, most are in < 1 seconds; > 2. China Unionpay and CPIC use Kylin to replace IBM Cognos cubes. One > Kylin cube replaced more than 100 Cognos cubes, with better building > performance and query performance. > 3. China Construction Bank uses Hadoop + Kylin to offload the Greenplum. > Some systems have been migrated to Kylin successfully. > 4. Yum (KFC) and several other users are using Kylin to replace Microsoft > SSAS. > 5. Meituan, Ctrip, JD, Didi, Xiao Mi, Huawei, OLX group, autohome.com.cn, > Xactly, and many others are using Kylin as the platform of their DaaS (Data > as a Service), providing data service to their thousands of internal > analysts and tens of thousands of external tenants. > > Now let's look at the definition of Data warehouse [1]: > > "*A data warehouse is a subject-oriented, integrated, time-variant and > non-volatile collection of data in support of management's decision-making > process.*" > > In Kylin, each model/cube is created for a certain subject; Kylin > integrates well with Hive, Hadoop, Spark, Kafka, and other systems; Kylin > incremental loads the data by time, build the cube and then save as > segments (partitions), and they are non-volatile unless you refresh them; > During the analysis (roll-up, drill-down, etc), the data is always > consistent. Kylin provides SQL interface and JDBC/ODBC/HTTP API for you to > easily connect from BI/visualization tools like Tableau and others. > > All in all, you can see that users are using Kylin not just as a SQL > engine, but also as an Analytical Data Warehouse, for very large scale data > (PB scale). In the world of big data, Kylin is unique. Its design is > elegant, its architecture is scalable and pluggable. In order to give > Kylin more visibility and can be discovered by more people, I propose to > change Kylin's position/slogan from the "OLAP engine for big data" to > "Analytical Data warehouse for big data". > > Please feel free to share your comments. > > [1] > https://www.1keydata.com/datawarehousing/data-warehouse-definition.html > > Best regards, > > Shaofeng Shi 史少锋 > Apache Kylin PMC > Email: shaofeng...@apache.org > > Apache Kylin FAQ: https://kylin.apache.org/docs/gettingstarted/faq.html > Join Kylin user mail group: user-subscr...@kylin.apache.org > Join Kylin dev mail group: dev-subscr...@kylin.apache.org > > > -- --------------------- Best regards, Ni Chunen / George