Drill Storage and Format Plugin Support Matrix
You can deploy Drill without Hadoop in a standalone configuration on a single node, however multi-node standalone cluster deployments of Drill are not supported. Note that Drill itself does not require Hadoop.
The following table lists the supported and unsupported data sources and formats in Drill:
Data Source | Storage Plugin Type | Formats | Supported |
file system | dfs | Text (CSV, TSV, PSV) | Yes |
Parquet | Yes | ||
JSON | Yes | ||
Avro | No | ||
HPE Ezmeral Data Fabric Database | dfs | Binary | Yes |
JSON | Yes | ||
HBase | hbase | Binary | No (as of Drill 1.11 and Core 6.0) |
Hive | hive | Text (CSV, TSV, PSV) | Yes |
Parquet | Yes | ||
JSON | Yes | ||
Avro | Yes | ||
Other Hive built-in SerDes | Yes (Not recommended due to the memory overhead and performance implications.) | ||
S3 | s3 | Supports the same formats as the dfs storage plugin. | Yes |
MongoDB | mongodb | N/A | No |
RDBMS | jdbc | N/A | No |
Kudu | kudu | N/A | No |
Kafka | kafka | JSON | No NOTE The kafka storage plugin on the Streams is in
the Alpha testing phase and not officially supported. See Configuring the Kafka Storage
Plugin for more information. |
OpenTSDB | openTSDB | N/A |
NOTE The openTSDB storage plugin is not officially supported. See OpenTSDB Storage Plugin for more information.
|
NOTE As of the Core 6.0 and Drill 1.11,
HBase is no longer supported, therefore the communication path between Drill and HBase is also
not supported. If you have an hbase storage plugin configured in Drill, you should disable
it.