This topic describes some of the limitations of using Exchange 3.x.
The correspondence between the NebulaGraph Exchange release (the JAR version) and the NebulaGraph core release is as follows.
JAR packages are available in two ways: compile them yourself or download them from the Maven repository.
If you are using NebulaGraph 1.x, use NebulaGraph Exchange 1.x.
Exchange 3.x supports the following operating systems:
- CentOS 7
To ensure the healthy operation of Exchange, ensure that the following software has been installed on the machine:
- Java version 1.8
- Scala version 2.10.7, 2.11.12, or 2.12.10
Apache Spark. The requirements for Spark versions when using Exchange to export data from data sources are as follows. In the following table, Y means that the corresponding Spark version is supported, and N means not supported.
Use the correct Exchange JAR file based on the Spark version. For example, for Spark version 2.4, use nebula-exchange_spark_2.4-3.0.0.jar.
Data source Spark 2.2 Spark 2.4 Spark 3 CSV file Y N Y JSON file Y Y Y ORC file Y Y Y Parquet file Y Y Y HBase Y Y Y MySQL Y Y Y PostgreSQL Y Y Y ClickHouse Y Y Y Neo4j N Y N Hive Y Y Y MaxCompute N Y N Pulsar N Y Untested Kafka N Y Untested NebulaGraph N Y N
Hadoop Distributed File System (HDFS) needs to be deployed in the following scenarios:
- Migrate HDFS data
- Generate SST files