This page describes how to integrate Delta Lake with Hive 4.x.x in ODP. The integration uses the ODP-provided Delta-Hive connector to query Delta tables created by Spark from Hive.
The Delta-Hive connector provides read-only access to Delta tables from Hive.
Download the Delta-Hive assembly connector JAR that corresponds to your ODP version.
Use the Delta-Hive connector provided for your ODP release. The upstream delta-hive_2.12:3.3.0 connector targets Hive 3.x and isn't compatible with Hive 4.1 in ODP 3.3.6.5.
Copy the JAR: Place the delta-hive-assembly_2.12-3.3.1.3.3.6.5-<build>.jar into the Hive lib directory. For example, on ODP 3.3.6.5-1012:
sudo cp delta-hive-assembly_2.12-3.3.1.3.3.6.5-1012.jar \
/usr/odp/3.3.6.5-1012/hive/lib/
Add the connector JAR to HIVE_AUX_JARS_PATH in hive-env.sh:
export HIVE_AUX_JARS_PATH=${HIVE_AUX_JARS_PATH}:/usr/odp/3.3.6.5-1012/hive/lib/delta-hive-assembly_2.12-3.3.1.3.3.6.5-1012.jar
Restart HiveServer2 from Ambari: Restart the Hive service (e.g., from Ambari) to apply the changes.
// Using Spark Shell or Scala
import org.apache.spark.sql.SparkSession
val spark = SparkSession.builder()
.appName("Create Delta Table")
.config("spark.sql.extensions", "io.delta.sql.DeltaSparkSessionExtension")
.config("spark.sql.catalog.spark_catalog", "org.apache.spark.sql.delta.catalog.DeltaCatalog")
.getOrCreate()
// Create sample data
val data = Seq(
(1L, "uuid1", "rider1", "driver1", 10.5, "city1"),
(2L, "uuid2", "rider2", "driver2", 20.5, "city2")
)
val df = data.toDF("ts", "uuid", "rider", "driver", "fare", "city")
df.write.format("delta").save("hdfs:///warehouse/tablespace/external/hive/my_delta_table")
Open Hive Shell:
Create the External Delta Table:
CREATE EXTERNAL TABLE deltaTableTest(
ts BIGINT,
uuid STRING,
rider STRING,
driver STRING,
fare DOUBLE,
city STRING
)
STORED BY 'io.delta.hive.DeltaStorageHandler'
LOCATION 'hdfs:///warehouse/tablespace/external/hive/my_delta_table';
Info
In case table is partitioned don’t add partitioned by clause, just add partition column as regular column.
The Delta Hive connector reads the partition information from the Delta transaction log (_delta_log) and handles partitioning transparently. This is by design to ensure consistency between the Hive table definition and the underlying Delta metadata.
After running the CREATE EXTERNAL TABLE command, you should see a success message indicating that the table has been created.
You can also verify with:
0: jdbc:hive2://oraupg03.acceldata.ce:2181,or> DESCRIBE FORMATTED deltaTableTest;
INFO : Compiling command(queryId=hive_20260202023428_624c2a02-26c0-4a19-9e5b-5ea846a5c6c2): DESCRIBE FORMATTED deltaTableTest
INFO : Semantic Analysis Completed (retrial = false)
INFO : Created Hive schema: Schema(fieldSchemas:[FieldSchema(name:col_name, type:string, comment:from deserializer), FieldSchema(name:data_type, type:string, comment:from deserializer), FieldSchema(name:comment, type:string, comment:from deserializer)], properties:null)
INFO : Completed compiling command(queryId=hive_20260202023428_624c2a02-26c0-4a19-9e5b-5ea846a5c6c2); Time taken: 0.164 seconds
INFO : Operation DESCTABLE obtained 0 locks
INFO : Executing command(queryId=hive_20260202023428_624c2a02-26c0-4a19-9e5b-5ea846a5c6c2): DESCRIBE FORMATTED deltaTableTest
INFO : Starting task [Stage-0:DDL] in serial mode
INFO : Completed executing command(queryId=hive_20260202023428_624c2a02-26c0-4a19-9e5b-5ea846a5c6c2); Time taken: 0.156 seconds
+-------------------------------+----------------------------------------------------+----------------------------------------------------+
| col_name | data_type | comment |
+-------------------------------+----------------------------------------------------+----------------------------------------------------+
| ts | bigint | |
| uuid | string | |
| rider | string | |
| driver | string | |
| fare | double | |
| city | string | |
| | NULL | NULL |
| # Detailed Table Information | NULL | NULL |
| Database: | default | NULL |
| OwnerType: | USER | NULL |
| Owner: | hive | NULL |
| CreateTime: | Mon Feb 02 02:34:16 IST 2026 | NULL |
| LastAccessTime: | UNKNOWN | NULL |
| Retention: | 0 | NULL |
| Location: | hdfs://nameservice1/warehouse/tablespace/external/hive/my_delta_table | NULL |
| Table Type: | EXTERNAL_TABLE | NULL |
| Table Parameters: | NULL | NULL |
| | EXTERNAL | TRUE |
| | bucketing_version | 2 |
| | numFiles | 2 |
| | spark.sql.sources.provider | DELTA |
| | storage_handler | io.delta.hive.DeltaStorageHandler |
| | totalSize | 3662 |
| | transient_lastDdlTime | 1769979856 |
| | NULL | NULL |
| # Storage Information | NULL | NULL |
| SerDe Library: | org.apache.hadoop.hive.ql.io.parquet.serde.ParquetHiveSerDe | NULL |
| InputFormat: | null | NULL |
| OutputFormat: | null | NULL |
| Compressed: | No | NULL |
| Num Buckets: | -1 | NULL |
| Bucket Columns: | [] | NULL |
| Sort Columns: | [] | NULL |
| Storage Desc Params: | NULL | NULL |
| | path | hdfs://nameservice1/warehouse/tablespace/external/hive/my_delta_table |
| | serialization.format | 1 |
+-------------------------------+----------------------------------------------------+----------------------------------------------------+
36 rows selected (0.421 seconds)
--To Read data from Hive set tez.input.format
0: jdbc:hive2://oraupg03.acceldata.ce:2181,or> SET hive.tez.input.format=io.delta.hive.HiveInputFormat;
No rows affected (0.012 seconds)
0: jdbc:hive2://oraupg03.acceldata.ce:2181,or> SELECT * FROM deltaTableTest LIMIT 10;
INFO : Compiling command(queryId=hive_20260202023459_d89b9135-9e0e-4e1d-941b-bf2f4a0fdac1): SELECT * FROM deltaTableTest LIMIT 10
INFO : No Stats for default@deltatabletest, Columns: fare, driver, city, uuid, rider, ts
INFO : Semantic Analysis Completed (retrial = false)
INFO : Created Hive schema: Schema(fieldSchemas:[FieldSchema(name:deltatabletest.ts, type:bigint, comment:null), FieldSchema(name:deltatabletest.uuid, type:string, comment:null), FieldSchema(name:deltatabletest.rider, type:string, comment:null), FieldSchema(name:deltatabletest.driver, type:string, comment:null), FieldSchema(name:deltatabletest.fare, type:double, comment:null), FieldSchema(name:deltatabletest.city, type:string, comment:null)], properties:null)
INFO : Completed compiling command(queryId=hive_20260202023459_d89b9135-9e0e-4e1d-941b-bf2f4a0fdac1); Time taken: 0.159 seconds
INFO : Operation QUERY obtained 0 locks
INFO : Executing command(queryId=hive_20260202023459_d89b9135-9e0e-4e1d-941b-bf2f4a0fdac1): SELECT * FROM deltaTableTest LIMIT 10
INFO : Completed executing command(queryId=hive_20260202023459_d89b9135-9e0e-4e1d-941b-bf2f4a0fdac1); Time taken: 0.004 seconds
+--------------------+----------------------+-----------------------+------------------------+----------------------+----------------------+
| deltatabletest.ts | deltatabletest.uuid | deltatabletest.rider | deltatabletest.driver | deltatabletest.fare | deltatabletest.city |
+--------------------+----------------------+-----------------------+------------------------+----------------------+----------------------+
| 2 | uuid2 | rider2 | driver2 | 20.5 | city2 |
| 1 | uuid1 | rider1 | driver1 | 10.5 | city1 |
+--------------------+----------------------+-----------------------+------------------------+----------------------+----------------------+
SELECT * FROM deltaTableTest LIMIT 10;
A single Delta-Hive connector deployment can read Delta tables written by the supported Spark versions.
Spark Version | Delta Version | Connector Support |
|---|
Spark 3.3.3 | Delta 2.3.0 | Yes |
Spark 3.5.1 | Delta 3.2.1 | Yes |
Spark 3.5.5 | Delta 3.3.3 | Yes |
Spark 4.1.1 | Delta 4.3.1 | Yes |
The Delta-Hive connector included with ODP 3.3.6.5 is built and validated with JDK 11 to match the Hive 4.1 runtime included with ODP. If you build the connector from source, use JDK 11.
Writing data to Delta tables directly from Hive isn't supported. Use Spark for INSERT, UPDATE, DELETE, MERGE, and other write operations.
The Delta-Hive connector provides read-only access to Delta tables from Hive.
Info
Writing to a Delta table directly through Hive is not supported. You must use Spark for data insertion.
Attempting to insert data directly via Hive will result in an error:
INSERT INTO deltaTableTest VALUES
(1693400000000, 'uuid1', 'rider1', 'driver1', 15.5, 'city1'),
(1693400001000, 'uuid2', 'rider2', 'driver2', 20.0, 'city2');
Error Message:
ERROR : FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.tez.TezTask. Writing to a Delta table in Hive is not supported. Please use Spark to write.
INFO : Completed executing command(queryId=hive_20260202023700_6d89c9ea-ea02-463b-be16-e7d8164a9328); Time taken: 0.152 seconds
Error: Error while compiling statement: FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.tez.TezTask. Writing to a Delta table in Hive is not supported. Please use Spark to write. (state=08S01,code=1)
As part of the Delta Lake integration (ODP-1697), we do support querying Delta tables via Hive using the internal Delta–Hive connector JAR, specifically built for Hive 4 compatibility in ODP 3.3.6.x. However, note the following key limitations:
Read-only support: Only SELECT and CREATE EXTERNAL TABLE operations are supported from Hive.
No write support: INSERT, UPDATE, DELETE, MERGE, or any DML operations on Delta tables from Hive are not supported, by design. This is enforced by the DeltaOutputFormat class in the connector, which throws UnsupportedOperationException for any write attempt.
Only EXTERNAL tables are supported: The Delta table must be pre-created using Spark with proper transaction logs (_delta_log), and then referenced in Hive using:
STORED BY 'io.delta.hive.DeltaStorageHandler'
Operation | Hive Support |
|---|
CREATE EXTERNAL TABLE (pointing to existing Delta)
| Yes |
SELECT / READ
| Yes |
INSERT | No |
UPDATE | No |
DELETE | No |
MERGE | No |
