E-MapReduce:Kyuubi

Features

• Multi-tenancy: Kyuubi provides end-to-end multi-tenancy support for resource acquisition, data access, and metadata access by using a unified authentication and authorization layer.

• High availability: Kyuubi uses ZooKeeper for load balancing, delivering enterprise-grade high availability and unlimited client concurrency.

• Multiple workloads: Kyuubi handles multiple workloads on a single platform with a single data replica and a unified SQL interface.

Scenarios

• Interactive analysis: Kyuubi enables enterprise-grade analytics platforms for interactive visual analytics on big data. It supports common computing frameworks along with Java Database Connectivity (JDBC) and Open Database Connectivity (ODBC) interfaces, allowing efficient querying through SQL or business intelligence (BI) tools. Kyuubi caches back-end engine instances at the user level to improve resource sharing and accelerate responses. It processes queries on large datasets in parallel and returns results quickly.

• Batch data processing: Kyuubi provides a SQL interface for common batch processing such as large extract, transform, and load (ETL) processes. Kyuubi and its engines are storage-independent and support numerous data sources. Kyuubi can also isolate back-end engine instances at the connection level for better resource isolation and stability.

Comparison of Kyuubi, Livy, and Spark ThriftServer

References

• The Share Level of Kyuubi Engines

• The TTL of Kyuubi Engines