This topic describes some of the limitations of using Exchange 3.x.
The correspondence between the Nebula Exchange release (the JAR version) and the Nebula Graph core release is as follows.
|Exchange client||Nebula Graph|
JAR packages are available in two ways: compile them yourself or download them from the Maven repository.
If you are using Nebula Graph 1.x, use Nebula 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 Nebula Graph N Y N
Hadoop Distributed File System (HDFS) needs to be deployed in the following scenarios:
- Migrate HDFS data
- Generate SST files