Iceberg Connector

Overview

The Iceberg connector allows querying data stored in Iceberg tables.

Metastores

Iceberg tables store most of the metadata in the metadata files, along with the data on the filesystem, but it still requires a central place to find the current location of the current metadata pointer for a table. This central place is called the Iceberg Catalog. The Presto Iceberg connector supports different types of Iceberg Catalogs : HIVE, NESSIE, REST, and HADOOP.

To configure the Iceberg connector, create a catalog properties file etc/catalog/iceberg.properties. To define the catalog type, iceberg.catalog.type property is required along with the following contents, with the property values replaced as follows:

Hive Metastore catalog

The Iceberg connector supports the same configuration for HMS as a Hive connector.

1. connector.name=iceberg
2. hive.metastore.uri=hostname:port
3. iceberg.catalog.type=hive

Glue catalog

The Iceberg connector supports the same configuration for Glue as a Hive connector.

1. connector.name=iceberg
2. hive.metastore=glue
3. iceberg.catalog.type=hive

There are additional configurations available when using the Iceberg connector configured with Hive or Glue catalogs.

Nessie catalog

To use a Nessie catalog, configure the catalog type as iceberg.catalog.type=nessie.

1. connector.name=iceberg
2. iceberg.catalog.type=nessie
3. iceberg.catalog.warehouse=/tmp
4. iceberg.nessie.uri=https://localhost:19120/api/v1

Additional supported properties for the Nessie catalog:

Setting Up Nessie With Docker

To set up a Nessie instance locally using the Docker image, see Setting up Nessie. Once the Docker instance is up and running, you should see logs similar to the following example:

1. 2023-09-05 13:11:37,905 INFO  [io.quarkus] (main) nessie-quarkus 0.69.0 on JVM (powered by Quarkus 3.2.4.Final) started in 1.921s. Listening on: http://0.0.0.0:19120
2. 2023-09-05 13:11:37,906 INFO  [io.quarkus] (main) Profile prod activated.
3. 2023-09-05 13:11:37,906 INFO  [io.quarkus] (main) Installed features: [agroal, amazon-dynamodb, cassandra-client, cdi, google-cloud-bigtable, hibernate-validator, jdbc-postgresql, logging-sentry, micrometer, mongodb-client, narayana-jta, oidc, opentelemetry, reactive-routes, resteasy, resteasy-jackson, security, security-properties-file, smallrye-context-propagation, smallrye-health, smallrye-openapi, swagger-ui, vertx]

If log messages related to Nessie’s OpenTelemetry collector appear similar to the following example, you can disable OpenTelemetry using the configuration option quarkus.otel.sdk.disabled=true.

1. 2023-08-27 11:10:02,492 INFO  [io.qua.htt.access-log] (executor-thread-1) 172.17.0.1 - - [27/Aug/2023:11:10:02 +0000] "GET /api/v1/config HTTP/1.1" 200 62
2. 2023-08-27 11:10:05,007 SEVERE [io.ope.exp.int.grp.OkHttpGrpcExporter] (OkHttp http://localhost:4317/...) Failed to export spans. The request could not be executed. Full error message: Failed to connect to localhost/127.0.0.1:4317

For example, start the Docker image using the following command: docker run -p 19120:19120 -e QUARKUS_OTEL_SDK_DISABLED=true ghcr.io/projectnessie/nessie

For more information about this configuration option and other related options, see the OpenTelemetry Configuration Reference.

For more information about troubleshooting OpenTelemetry traces, see Troubleshooting traces.

If an error similar to the following example is displayed, this is probably because you are interacting with an http server, and not an https server. You need to set iceberg.nessie.uri to http://localhost:19120/api/v1.

1. Caused by: javax.net.ssl.SSLException: Unsupported or unrecognized SSL message
2. at sun.security.ssl.SSLSocketInputRecord.handleUnknownRecord(SSLSocketInputRecord.java:448)
3. at sun.security.ssl.SSLSocketInputRecord.decode(SSLSocketInputRecord.java:174)
4. at sun.security.ssl.SSLTransport.decode(SSLTransport.java:111)
5. at sun.security.ssl.SSLSocketImpl.decode(SSLSocketImpl.java:1320)
6. at sun.security.ssl.SSLSocketImpl.readHandshakeRecord(SSLSocketImpl.java:1233)
7. at sun.security.ssl.SSLSocketImpl.startHandshake(SSLSocketImpl.java:417)
8. at sun.security.ssl.SSLSocketImpl.startHandshake(SSLSocketImpl.java:389)
9. at sun.net.www.protocol.https.HttpsClient.afterConnect(HttpsClient.java:558)
10. at sun.net.www.protocol.https.AbstractDelegateHttpsURLConnection.connect(AbstractDelegateHttpsURLConnection.java:201)
11. at sun.net.www.protocol.https.HttpsURLConnectionImpl.connect(HttpsURLConnectionImpl.java:167)
12. at org.projectnessie.client.http.impl.jdk8.UrlConnectionRequest.executeRequest(UrlConnectionRequest.java:71)
13. ... 42 more

REST catalog

To use a REST catalog, configure the catalog type as iceberg.catalog.type=rest. A minimal configuration includes:

1. connector.name=iceberg
2. iceberg.catalog.type=rest
3. iceberg.rest.uri=https://localhost:8181

Additional supported properties for the REST catalog:

Hadoop catalog

To use a Hadoop catalog, configure the catalog type as iceberg.catalog.type=hadoop. A minimal configuration includes:

1. connector.name=iceberg
2. iceberg.catalog.type=hadoop
3. iceberg.catalog.warehouse=hdfs://hostname:port

Hadoop catalog configuration properties:

Configuration Properties

Note

The Iceberg connector supports configuration options for Amazon S3 as a Hive connector.

The following configuration properties are available for all catalog types:

Table Properties

Table properties set metadata for the underlying tables. This is key for CREATE TABLE/CREATE TABLE AS statements. Table properties are passed to the connector using a WITH clause:

1. CREATE TABLE tablename
2. WITH (
3. property_name = property_value,
4. ...
5. )

The following table properties are available, which are specific to the Presto Iceberg connector:

The table definition below specifies format ORC, partitioning by columns c1 and c2, and a file system location of s3://test_bucket/test_schema/test_table:

1. CREATE TABLE test_table (
2. c1 bigint,
3. c2 varchar,
4. c3 double
5. )
6. WITH (
7. format = 'ORC',
8. partitioning = ARRAY['c1', 'c2'],
9. location = 's3://test_bucket/test_schema/test_table')
10. )

Session Properties

Session properties set behavior changes for queries executed within the given session.

Caching Support

Manifest File Caching

As of Iceberg version 1.1.0, Apache Iceberg provides a mechanism to cache the contents of Iceberg manifest files in memory. This feature helps to reduce repeated reads of small Iceberg manifest files from remote storage.

Note

Currently, manifest file caching is supported for Hadoop and Nessie catalogs in the Presto Iceberg connector.

The following configuration properties are available:

Alluxio Data Cache

A Presto worker caches remote storage data in its original form (compressed and possibly encrypted) on local SSD upon read.

The following configuration properties are required to set in the Iceberg catalog file (catalog/iceberg.properties):

1. cache.enabled=true
2. cache.base-directory=file:///mnt/flash/data
3. cache.type=ALLUXIO
4. cache.alluxio.max-cache-size=1600GB
5. hive.node-selection-strategy=SOFT_AFFINITY

JMX queries to get the metrics and verify the cache usage:

1. SELECT * FROM jmx.current."com.facebook.alluxio:name=client.cachehitrate,type=gauges";

3. SELECT * FROM jmx.current."com.facebook.alluxio:name=client.cachebytesreadcache,type=meters";

5. SHOW TABLES FROM jmx.current like '%alluxio%';

File And Stripe Footer Cache

Caches open file descriptors and stripe or file footer information in leaf worker memory. These pieces of data are mostly frequently accessed when reading files.

The following configuration properties are required to set in the Iceberg catalog file (catalog/iceberg.properties):

1. # scheduling
2. hive.node-selection-strategy=SOFT_AFFINITY

4. # orc
5. iceberg.orc.file-tail-cache-enabled=true
6. iceberg.orc.file-tail-cache-size=100MB
7. iceberg.orc.file-tail-cache-ttl-since-last-access=6h
8. iceberg.orc.stripe-metadata-cache-enabled=true
9. iceberg.orc.stripe-footer-cache-size=100MB
10. iceberg.orc.stripe-footer-cache-ttl-since-last-access=6h
11. iceberg.orc.stripe-stream-cache-size=300MB
12. iceberg.orc.stripe-stream-cache-ttl-since-last-access=6h

14. # parquet
15. iceberg.parquet.metadata-cache-enabled=true
16. iceberg.parquet.metadata-cache-size=100MB
17. iceberg.parquet.metadata-cache-ttl-since-last-access=6h

JMX queries to get the metrics and verify the cache usage:

1. SELECT * FROM jmx.current."com.facebook.presto.hive:name=iceberg_parquetmetadata,type=cachestatsmbean";

Metastore Cache

Metastore Cache only caches the schema, table, and table statistics. The table object cached in the tableCache is only used for reading the table metadata location and table properties and, the rest of the table metadata is fetched from the filesystem/object storage metadata location.

Note

Metastore Cache would be applicable only for Hive Catalog in the Presto Iceberg connector.

1. hive.metastore-cache-ttl=2d
2. hive.metastore-refresh-interval=3d
3. hive.metastore-cache-maximum-size=10000000

Extra Hidden Metadata Columns

The Iceberg connector exposes extra hidden metadata columns. You can query these as part of a SQL query by including them in your SELECT statement.

$path column

• $path: Full file system path name of the file for this row

1. SELECT "$path", regionkey FROM "ctas_nation";

1. $path                    |  regionkey
2. ---------------------------------+-----------
3. /full/path/to/file/file.parquet | 2

$data_sequence_number column

• $data_sequence_number: The Iceberg data sequence number in which this row was added

1. SELECT "$data_sequence_number", regionkey FROM "ctas_nation";

1. $data_sequence_number     |  regionkey
2. ----------------------------------+------------
3. 2                    | 3

Extra Hidden Metadata Tables

The Iceberg connector exposes extra hidden metadata tables. You can query these as a part of a SQL query by appending name to the table.

$properties Table

• $properties : General properties of the given table

1. SELECT * FROM "ctas_nation$properties";

1. key           |  value
2. ----------------------+---------
3. write.format.default | PARQUET

$history Table

• $history : History of table state changes

1. SELECT * FROM "ctas_nation$history";

1. made_current_at            |     snapshot_id     | parent_id | is_current_ancestor
2. --------------------------------------+---------------------+-----------+---------------------
3. 2022-11-25 20:56:31.784 Asia/Kolkata  | 7606232158543069775 | NULL      | true

$snapshots Table

• $snapshots : Details about the table snapshots. For more information see Snapshots in the Iceberg Table Spec.

1. SELECT * FROM "ctas_nation$snapshots";

1. committed_at             |     snapshot_id     | parent_id | operation |                                                  manifest_list                                           |                                                                                 summary
2. --------------------------------------+---------------------+-----------+-----------+----------------------------------------------------------------------------------------------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
3. 2022-11-25 20:56:31.784 Asia/Kolkata  | 7606232158543069775 | NULL      | append    | s3://my-bucket/ctas_nation/metadata/snap-7606232158543069775-1-395a2cad-b244-409b-b030-cc44949e5a4e.avro | {changed-partition-count=1, added-data-files=1, total-equality-deletes=0, added-records=25, total-position-deletes=0, added-files-size=1648, total-delete-files=0, total-files-size=1648, total-records=25, total-data-files=1}

$manifests Table

• $manifests : Details about the manifests of different table snapshots. For more information see Manifests in the Iceberg Table Spec.

1. SELECT * FROM "ctas_nation$manifests";

1. path                                  | length | partition_spec_id |  added_snapshot_id  | added_data_files_count | existing_data_files_count | deleted_data_files_count | partitions
2. ---------------------------------------------------------------------------------+--------+-------------------+---------------------+------------------------+---------------------------+--------------------------+-----------
3. s3://my-bucket/ctas_nation/metadata/395a2cad-b244-409b-b030-cc44949e5a4e-m0.avro |   5957 |                 0 | 7606232158543069775 |                      1 |                         0 |                    0     |    []

$partitions Table

• $partitions : Detailed partition information for the table

1. SELECT * FROM "ctas_nation$partitions";

1. row_count | file_count | total_size |           nationkey           |                   name                   |          regionkey           |                          comment
2. -----------+------------+------------+-------------------------------+------------------------------------------+------------------------------+------------------------------------------------------------
3. 25     |          1 |       1648 | {min=0, max=24, null_count=0} | {min=ALGERIA, max=VIETNAM, null_count=0} | {min=0, max=4, null_count=0} | {min= haggle. careful, max=y final packaget, null_count=0}

$files Table

• $files : Overview of data files in the current snapshot of the table

1. SELECT * FROM "ctas_nation$files";

1. content |                                      file_path                               | file_format | record_count | file_size_in_bytes |        column_sizes         |       value_counts       |  null_value_counts   | nan_value_counts |          lower_bounds                     |             upper_bounds                   | key_metadata | split_offsets | equality_ids
2. ---------+------------------------------------------------------------------------------+-------------+--------------+--------------------+-----------------------------+--------------------------+----------------------+------------------+-------------------------------------------+--------------------------------------------+--------------+---------------+-------------
3. 0     | s3://my-bucket/ctas_nation/data/9f889274-6f74-4d28-8164-275eef99f660.parquet | PARQUET     |           25 |               1648 | {1=52, 2=222, 3=105, 4=757} | {1=25, 2=25, 3=25, 4=25} | {1=0, 2=0, 3=0, 4=0} |  NULL            | {1=0, 2=ALGERIA, 3=0, 4= haggle. careful} | {1=24, 2=VIETNAM, 3=4, 4=y final packaget} | NULL         | NULL          | NULL

$changelog Table

This table lets you view which row-level changes have occurred to the table in a particular order over time. The $changelog table represents the history of changes to the table, while also making the data available to process through a query.

The result of a changelog query always returns a static schema with four columns:

1. operation: (VARCHAR) indicating whether the row was inserted, updated, or deleted.
2. ordinal: (int) A number indicating a relative order that a particular change needs to be applied to the table relative to all other changes.
3. snapshotid: (bigint) Represents the snapshot a row-level change was made in.
4. rowdata: (row(T)) which includes the data for the particular row. The inner values of this type match the schema of the parent table.

The changelog table can be queried with the following name format:

1. ... FROM "<table>[@<begin snapshot ID>]$changelog[@<end snapshot ID>]"

• <table> is the name of the table.
• <begin snapshot ID> is the snapshot of the table you want to begin viewing changes from. This parameter is optional. If absent, the oldest available snapshot is used.
• <end snapshot ID> is the last snapshot for which you want to view changes. This parameter is optional. If absent, the most current snapshot of the table is used.

One use for the $changelog table would be to find when a record was inserted or removed from the table. To accomplish this, the $changelog table can be used in conjunction with the $snapshots table. First, choose a snapshot ID from the $snapshots table to choose the starting point.

1. SELECT * FROM "orders$snapshots";

1. committed_at                 |     snapshot_id     |      parent_id      | operation |                                                                                       manifest_list                                                                                        |                                                                                                              summary
2. ---------------------------------------------+---------------------+---------------------+-----------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
3. 2023-09-26 08:45:20.930 America/Los_Angeles | 2423571386296047175 | NULL                | append    | file:/var/folders/g_/6_hxl7r16qdddw7956j_r88h0000gn/T/PrestoTest8140889264166671718/catalog/tpch/ctas_orders/metadata/snap-2423571386296047175-1-3f288b1c-95a9-406b-9e17-9cfe31a11b48.avro | {changed-partition-count=1, added-data-files=4, total-equality-deletes=0, added-records=100, total-position-deletes=0, added-files-size=9580, total-delete-files=0, total-files-size=9580, total-records=100, total-data-files=4}
4. 2023-09-26 08:45:36.942 America/Los_Angeles | 8702997868627997320 | 2423571386296047175 | append    | file:/var/folders/g_/6_hxl7r16qdddw7956j_r88h0000gn/T/PrestoTest8140889264166671718/catalog/tpch/ctas_orders/metadata/snap-8702997868627997320-1-a2e1c714-7eed-4e2c-b144-dae4147ebaa4.avro | {changed-partition-count=1, added-data-files=1, total-equality-deletes=0, added-records=1, total-position-deletes=0, added-files-size=1687, total-delete-files=0, total-files-size=11267, total-records=101, total-data-files=5}
5. 2023-09-26 08:45:39.866 America/Los_Angeles | 7615903782581283889 | 8702997868627997320 | append    | file:/var/folders/g_/6_hxl7r16qdddw7956j_r88h0000gn/T/PrestoTest8140889264166671718/catalog/tpch/ctas_orders/metadata/snap-7615903782581283889-1-d94c2114-fd22-4de2-9ab5-c0b5bf67282f.avro | {changed-partition-count=1, added-data-files=3, total-equality-deletes=0, added-records=3, total-position-deletes=0, added-files-size=4845, total-delete-files=0, total-files-size=16112, total-records=104, total-data-files=8}
6. 2023-09-26 08:45:48.404 America/Los_Angeles |  677209275408372885 | 7615903782581283889 | append    | file:/var/folders/g_/6_hxl7r16qdddw7956j_r88h0000gn/T/PrestoTest8140889264166671718/catalog/tpch/ctas_orders/metadata/snap-677209275408372885-1-ad69e208-1440-459b-93e8-48e61f961758.avro  | {changed-partition-count=1, added-data-files=3, total-equality-deletes=0, added-records=5, total-position-deletes=0, added-files-size=4669, total-delete-files=0, total-files-size=20781, total-records=109, total-data-files=11}

Now that we know the snapshots available to query in the changelog, we can see what changes were made to the table since it was created. Specifically, this example uses the earliest snapshot ID: 2423571386296047175

1. SELECT * FROM "ctas_orders@2423571386296047175$changelog" ORDER BY ordinal;

1. operation | ordinal |     snapshotid      |                                                                                                                   rowdata
2. -----------+---------+---------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
3. INSERT    |       0 | 8702997868627997320 | {orderkey=37504, custkey=1291, orderstatus=O, totalprice=165509.83, orderdate=1996-03-04, orderpriority=5-LOW, clerk=Clerk#000000871, shippriority=0, comment=c theodolites alongside of the fluffily bold requests haggle quickly against }
4. INSERT    |       1 | 7615903782581283889 | {orderkey=12001, custkey=739, orderstatus=F, totalprice=138635.75, orderdate=1994-07-07, orderpriority=2-HIGH, clerk=Clerk#000000863, shippriority=0, comment=old, even theodolites. regular, special theodolites use furio}
5. INSERT    |       1 | 7615903782581283889 | {orderkey=17989, custkey=364, orderstatus=F, totalprice=133669.05, orderdate=1994-01-17, orderpriority=4-NOT SPECIFIED, clerk=Clerk#000000547, shippriority=0, comment=ously express excuses. even theodolit}
6. INSERT    |       1 | 7615903782581283889 | {orderkey=37504, custkey=1291, orderstatus=O, totalprice=165509.83, orderdate=1996-03-04, orderpriority=5-LOW, clerk=Clerk#000000871, shippriority=0, comment=c theodolites alongside of the fluffily bold requests haggle quickly against }
7. INSERT    |       2 |  677209275408372885 | {orderkey=17991, custkey=92, orderstatus=O, totalprice=20732.51, orderdate=1998-07-09, orderpriority=4-NOT SPECIFIED, clerk=Clerk#000000636, shippriority=0, comment= the quickly express accounts. iron}
8. INSERT    |       2 |  677209275408372885 | {orderkey=17989, custkey=364, orderstatus=F, totalprice=133669.05, orderdate=1994-01-17, orderpriority=4-NOT SPECIFIED, clerk=Clerk#000000547, shippriority=0, comment=ously express excuses. even theodolit}
9. INSERT    |       2 |  677209275408372885 | {orderkey=17990, custkey=458, orderstatus=O, totalprice=218031.58, orderdate=1998-03-18, orderpriority=3-MEDIUM, clerk=Clerk#000000340, shippriority=0, comment=ounts wake final foxe}
10. INSERT    |       2 |  677209275408372885 | {orderkey=18016, custkey=403, orderstatus=O, totalprice=174070.99, orderdate=1996-03-19, orderpriority=1-URGENT, clerk=Clerk#000000629, shippriority=0, comment=ly. quickly ironic excuses are furiously. carefully ironic pack}
11. INSERT    |       2 |  677209275408372885 | {orderkey=18017, custkey=958, orderstatus=F, totalprice=203091.02, orderdate=1993-03-26, orderpriority=1-URGENT, clerk=Clerk#000000830, shippriority=0, comment=sleep quickly bold requests. slyly pending pinto beans haggle in pla}

Procedures

Use the CALL statement to perform data manipulation or administrative tasks. Procedures are available in the system schema of the catalog.

Register Table

Iceberg tables for which table data and metadata already exist in the file system can be registered with the catalog. Use the register_table procedure on the catalog’s system schema and supply the target schema, desired table name, and the location of the table metadata:

1. CALL iceberg.system.register_table('schema_name', 'table_name', 'hdfs://localhost:9000/path/to/iceberg/table/metadata/dir')

Note

If multiple metadata files of the same version exist at the specified location, the most recently modified one is used.

A metadata file can optionally be included as an argument to register_table where a specific metadata file contains the targeted table state:

1. CALL iceberg.system.register_table('schema_name', 'table_name', 'hdfs://localhost:9000/path/to/iceberg/table/metadata/dir', '00000-35a08aed-f4b0-4010-95d2-9d73ef4be01c.metadata.json')

Note

The Iceberg REST catalog may not support table register depending on the type of the backing catalog.

Note

When registering a table with the Hive metastore, the user calling the procedure is set as the owner of the table and has SELECT, INSERT, UPDATE, and DELETE privileges for that table. These privileges can be altered using the GRANT and REVOKE commands.

Note

When using the Hive catalog, attempts to read registered Iceberg tables using the Hive connector will fail.

Unregister Table

Iceberg tables can be unregistered from the catalog using the unregister_table procedure on the catalog’s system schema:

1. CALL iceberg.system.unregister_table('schema_name', 'table_name')

Note

Table data and metadata remain in the filesystem after a call to unregister_table only when using the Hive catalog. This is similar to the behavior listed for the DROP TABLE command.

Rollback to Snapshot

Roll back a table to a specific snapshot ID. Iceberg can roll back to a specific snapshot ID by using the rollback_to_snapshot procedure on Icebergssystem` schema:

1. CALL iceberg.system.rollback_to_snapshot('table_name', 'snapshot_id');

The following arguments are available:

Expire Snapshots

Each DML (Data Manipulation Language) action in Iceberg produces a new snapshot while keeping the old data and metadata for snapshot isolation and time travel. Use expire_snapshots to remove older snapshots and their files.

This procedure removes old snapshots and their corresponding files, and never removes files which are required by a non-expired snapshot.

The following arguments are available:

Examples:

• Remove snapshots older than a specific day and time, but retain the last 10 snapshots:

1. CALL iceberg.system.expire_snapshots('schema_name', 'table_name', TIMESTAMP '2023-08-31 00:00:00.000', 10); - Remove snapshots with snapshot ID 10001 and 10002 (note that these snapshot IDs should not be the current snapshot):

1. CALL iceberg.system.expire_snapshots(schema => 'schema_name', table_name => 'table_name', snapshot_ids => ARRAY[10001, 10002]);

Remove Orphan Files

Use to remove files which are not referenced in any metadata files of an Iceberg table.

The following arguments are available:

Examples:

• Remove any files which are not known to the table db.sample and older than specified timestamp:

1. CALL iceberg.system.remove_orphan_files('db', 'sample', TIMESTAMP '2023-08-31 00:00:00.000'); - Remove any files which are not known to the table db.sample and created 3 days ago (by default):

1. CALL iceberg.system.remove_orphan_files(schema => 'db', table_name => 'sample');

SQL Support

The Iceberg connector supports querying and manipulating Iceberg tables and schemas (databases). Here are some examples of the SQL operations supported by Presto:

CREATE SCHEMA

Create a new Iceberg schema named web that stores tables in an S3 bucket named my-bucket:

1. CREATE SCHEMA iceberg.web
2. WITH (location = 's3://my-bucket/')

CREATE TABLE

Create a new Iceberg table named page_views in the web schema that is stored using the ORC file format, partitioned by ds and country:

1. CREATE TABLE iceberg.web.page_views (
2. view_time timestamp,
3. user_id bigint,
4. page_url varchar,
5. ds date,
6. country varchar
7. )
8. WITH (
9. format = 'ORC',
10. partitioning = ARRAY['ds', 'country']
11. )

Create an Iceberg table with Iceberg format version 2:

1. CREATE TABLE iceberg.web.page_views_v2 (
2. view_time timestamp,
3. user_id bigint,
4. page_url varchar,
5. ds date,
6. country varchar
7. )
8. WITH (
9. format = 'ORC',
10. partitioning = ARRAY['ds', 'country'],
11. format_version = '2'
12. )

Partition Column Transform

Beyond selecting some particular columns for partitioning, you can use the transform functions and partition the table by the transformed value of the column.

Available transforms in the Presto Iceberg connector include:

• Bucket (partitions data into a specified number of buckets using a hash function)
• Truncate (partitions the table based on the truncated value of the field and can specify the width of the truncated value)
• Identity (partitions data using unmodified source value)
• Year (partitions data using integer value by extracting a date or timestamp year, as years from 1970)
• Month (partitions data using integer value by extracting a date or timestamp month, as months from 1970-01-01)
• Day (partitions data using integer value by extracting a date or timestamp day, as days from 1970-01-01)
• Hour (partitions data using integer value by extracting a timestamp hour, as hours from 1970-01-01 00:00:00)

Create an Iceberg table partitioned into 8 buckets of equal size ranges:

1. CREATE TABLE players (
2. id int,
3. name varchar,
4. team varchar
5. )
6. WITH (
7. format = 'ORC',
8. partitioning = ARRAY['bucket(team, 8)']
9. );

Create an Iceberg table partitioned by the first letter of the team field:

1. CREATE TABLE players (
2. id int,
3. name varchar,
4. team varchar
5. )
6. WITH (
7. format = 'ORC',
8. partitioning = ARRAY['truncate(team, 1)']
9. );

Create an Iceberg table partitioned by ds:

1. CREATE TABLE players (
2. id int,
3. name varchar,
4. team varchar,
5. ds date
6. )
7. WITH (
8. format = 'ORC',
9. partitioning = ARRAY['year(ds)']
10. );

Create an Iceberg table partitioned by ts:

1. CREATE TABLE players (
2. id int,
3. name varchar,
4. team varchar,
5. ts timestamp
6. )
7. WITH (
8. format = 'ORC',
9. partitioning = ARRAY['hour(ts)']
10. );

CREATE VIEW

The Iceberg connector supports creating views in Hive and Glue metastores. To create a view named view_page_views for the iceberg.web.page_views table created in the CREATE TABLE example:

1. CREATE VIEW iceberg.web.view_page_views AS SELECT user_id, country FROM iceberg.web.page_views;

INSERT INTO

Insert data into the page_views table:

1. INSERT INTO iceberg.web.page_views VALUES(TIMESTAMP '2023-08-12 03:04:05.321', 1, 'https://example.com', current_date, 'country');

CREATE TABLE AS SELECT

Create a new table page_views_new from an existing table page_views:

1. CREATE TABLE iceberg.web.page_views_new AS SELECT * FROM iceberg.web.page_views

SELECT

SELECT table operations are supported for Iceberg format version 1 and version 2 in the connector:

1. SELECT * FROM iceberg.web.page_views;

3. SELECT * FROM iceberg.web.page_views_v2;

Table with delete files

Iceberg V2 tables support row-level deletion. For more information see Row-level deletes in the Iceberg Table Spec. Presto supports reading delete files, including Position Delete Files and Equality Delete Files. When reading, Presto merges these delete files to read the latest results.

ALTER TABLE

Alter table operations are supported in the Iceberg connector:

1. ALTER TABLE iceberg.web.page_views ADD COLUMN zipcode VARCHAR;

3. ALTER TABLE iceberg.web.page_views RENAME COLUMN zipcode TO location;

5. ALTER TABLE iceberg.web.page_views DROP COLUMN location;

To add a new column as a partition column, identify the transform functions for the column. The table is partitioned by the transformed value of the column:

1. ALTER TABLE iceberg.web.page_views ADD COLUMN zipcode VARCHAR WITH (partitioning = 'identity');

3. ALTER TABLE iceberg.web.page_views ADD COLUMN location VARCHAR WITH (partitioning = 'truncate(2)');

5. ALTER TABLE iceberg.web.page_views ADD COLUMN location VARCHAR WITH (partitioning = 'bucket(8)');

7. ALTER TABLE iceberg.web.page_views ADD COLUMN dt date WITH (partitioning = 'year');

9. ALTER TABLE iceberg.web.page_views ADD COLUMN ts timestamp WITH (partitioning = 'month');

11. ALTER TABLE iceberg.web.page_views ADD COLUMN dt date WITH (partitioning = 'day');

13. ALTER TABLE iceberg.web.page_views ADD COLUMN ts timestamp WITH (partitioning = 'hour');

TRUNCATE

The Iceberg connector can delete all of the data from tables without dropping the table from the metadata catalog using TRUNCATE TABLE.

1. TRUNCATE TABLE nation;

1. TRUNCATE TABLE;

1. SELECT * FROM nation;

1. nationkey | name | regionkey | comment
2. -----------+------+-----------+---------
3. (0 rows)

DELETE

The Iceberg connector can delete data from tables by using DELETE FROM. For example, to delete from the table lineitem:

1. DELETE FROM lineitem;

3. DELETE FROM lineitem WHERE linenumber = 1;

5. DELETE FROM lineitem WHERE linenumber not in (1, 3, 5, 7) and linestatus in ('O', 'F');

Note

Filtered columns only support comparison operators, such as EQUALS, LESS THAN, or LESS THAN EQUALS.

Deletes must only occur on the latest snapshot.

For V1 tables, the Iceberg connector can only delete data in one or more entire partitions. Columns in the filter must all be identity transformed partition columns of the target table.

DROP TABLE

Drop the table page_views

1. DROP TABLE iceberg.web.page_views

• Dropping an Iceberg table with Hive Metastore and Glue catalogs only removes metadata from metastore.
• Dropping an Iceberg table with Hadoop and Nessie catalogs removes all the data and metadata in the table.

DROP VIEW

Drop the view view_page_views:

1. DROP VIEW iceberg.web.view_page_views;

DROP SCHEMA

Drop the schema iceberg.web:

1. DROP SCHEMA iceberg.web

SHOW CREATE TABLE

Show the SQL statement that creates the specified Iceberg table by using SHOW CREATE TABLE.

For example, SHOW CREATE TABLE from the partitioned Iceberg table customer:

1. SHOW CREATE TABLE customer;

``` 1. CREATE TABLE iceberg.tpch_iceberg.customer ( 2. "custkey" bigint, 3. "name" varchar, 4. "address" varchar, 5. "nationkey" bigint, 6. "phone" varchar, 7. "acctbal" double, 8. "mktsegment" varchar, 9. "comment" varchar 10. ) 11. WITH ( 12. delete_mode = 'copy-on-write', 13. form

（本文超出站点单文档体积上限，已截断）