TiDB is an open-source NewSQL database that supports Hybrid Transactional and Analytical Processing (HTAP) workloads.
TiSpark is a thin layer built for running Apache Spark on top of TiDB/TiKV to answer complex OLAP queries. It enjoys the merits of both the Spark platform and the distributed clusters of TiKV while seamlessly integrated to TiDB to provide one-stop HTAP solutions for online transactions and analyses.
This article assumes that you have mastered the basic knowledge and operation of TiDB and TiSpark. For the knowledge not mentioned in this article, you can obtain it from TiDB Official Documentation.
By using kyuubi, we can run SQL queries towards TiDB/TiKV which is more convenient, easy to understand, and easy to expand than directly using spark to manipulate TiDB/TiKV.
To enable the integration of kyuubi spark sql engine and TiDB through Apache Spark Datasource V2 and Catalog APIs, you need to:
The classpath of kyuubi spark sql engine with TiDB supported consists of
kyuubi-spark-sql-engine-1.9.0-SNAPSHOT_2.12.jar, the engine jar deployed with Kyuubi distributions
a copy of spark distribution
tispark-assembly-<spark.version>_<scala.version>-<tispark.version>.jar (example: tispark-assembly-3.2_2.12-3.0.1.jar), which can be found in the Maven Central
In order to make the TiSpark packages visible for the runtime classpath of engines, we can use one of these methods:
Put the TiSpark packages into
Please mind the compatibility of different TiDB, TiSpark and Spark versions, which can be confirmed on the page of TiSpark Environment setup.
To activate functionality of TiSpark, we can set the following configurations:
spark.tispark.pd.addresses $pd_host:$pd_port spark.sql.extensions org.apache.spark.sql.TiExtensions spark.sql.catalog.tidb_catalog org.apache.spark.sql.catalyst.catalog.TiCatalog spark.sql.catalog.tidb_catalog.pd.addresses $pd_host:$pd_port
The spark.tispark.pd.addresses and spark.sql.catalog.tidb_catalog.pd.addresses configurations allow you to put in multiple PD servers. Specify the port number for each of them.
For example, when you have multiple PD servers on 10.16.20.1,10.16.20.2,10.16.20.3 with the port 2379, put it as 10.16.20.1:2379,10.16.20.2:2379,10.16.20.3:2379.
SELECT as a example,
SELECT * FROM foo;
DELETE FROM as a example, Spark 3 added support for DELETE FROM queries to remove data from tables.
DELETE FROM foo WHERE id >= 1 and id < 2;
As for now (TiSpark 3.0.1), TiSpark does not support
INSERT INTO/OVERWRITE operations
through Apache Spark Datasource V2 and Catalog APIs.