Iceberg коннектор#
Примечание
Ниже приведена оригинальная документация Trino. Скоро мы ее переведем на русский язык и дополним полезными примерами.
Apache Iceberg is an open table format for huge analytic datasets. The Iceberg connector allows querying data stored in files written in Iceberg format, as defined in the Iceberg Table Spec. It supports Apache Iceberg table spec version 1 and 2.
The Iceberg table state is maintained in metadata files. All changes to table state create a new metadata file and replace the old metadata with an atomic swap. The table metadata file tracks the table schema, partitioning config, custom properties, and snapshots of the table contents.
Iceberg data files can be stored in either Parquet, ORC or Avro format, as
determined by the format property in the table definition. The
table format defaults to ORC.
Iceberg is designed to improve on the known scalability limitations of Hive, which stores table metadata in a metastore that is backed by a relational database such as MySQL. It tracks partition locations in the metastore, but not individual data files. Trino queries using the Hive коннектор must first call the metastore to get partition locations, then call the underlying filesystem to list all data files inside each partition, and then read metadata from each data file.
Since Iceberg stores the paths to data files in the metadata files, it only consults the underlying file system for files that must be read.
Требования#
To use Iceberg, you need:
Network access from the Trino coordinator and workers to the distributed object storage.
Access to a Hive metastore service (HMS) or AWS Glue.
Network access from the Trino coordinator to the HMS. Hive metastore access with the Thrift protocol defaults to using port 9083.
Конфигурация#
The connector supports multiple Iceberg catalog types, you may use either a Hive
metastore service (HMS), AWS Glue, or a REST catalog. The catalog type is determined by the
iceberg.catalog.type property, it can be set to HIVE_METASTORE, GLUE, JDBC, or REST.
Hive metastore catalog#
The Hive metastore catalog is the default implementation.
When using it, the Iceberg connector supports the same metastore
configuration properties as the Hive connector. At a minimum,
hive.metastore.uri must be configured, see
Thrift metastore configuration.
connector.name=iceberg
hive.metastore.uri=thrift://localhost:9083
Glue catalog#
When using the Glue catalog, the Iceberg connector supports the same configuration properties as the Hive connector’s Glue setup. See AWS Glue metastore configuration.
connector.name=iceberg
iceberg.catalog.type=glue
REST catalog#
In order to use the Iceberg REST catalog, ensure to configure the catalog type with
iceberg.catalog.type=rest and provide further details with the following
properties:
Property Name |
Description |
|---|---|
|
REST server API endpoint URI (required).
Example: |
|
The type of security to use (default: |
|
Session information included when communicating with the REST Catalog.
Options are |
|
The Bearer token which will be used for interactions
with the server. A |
|
The credential to exchange for a token in the OAuth2 client
credentials flow with the server. A |
connector.name=iceberg
iceberg.catalog.type=rest
iceberg.rest-catalog.uri=http://iceberg-with-rest:8181
JDBC catalog#
Предупреждение
The JDBC catalog may face the compatibility issue if Iceberg introduces breaking changes in the future. Consider the REST catalog as an alternative solution.
At a minimum, iceberg.jdbc-catalog.connection-url and
iceberg.jdbc-catalog.catalog-name must be configured.
When using any database besides PostgreSQL, a JDBC driver jar file must be placed in the plugin directory.
connector.name=iceberg
iceberg.catalog.type=jdbc
iceberg.jdbc-catalog.catalog-name=test
iceberg.jdbc-catalog.connection-url=jdbc:postgresql://example.net:5432/database?user=admin&password=test
iceberg.jdbc-catalog.default-warehouse-dir=s3://bucket
General configuration#
These configuration properties are independent of which catalog implementation is used.
Property name |
Description |
Default |
|---|---|---|
|
Define the data storage file format for Iceberg tables. Possible values are
|
|
|
The compression codec to be used when writing files. Possible values are
|
|
|
Read file sizes from metadata instead of file system. This property should only be set as a workaround for this issue. The problem was fixed in Iceberg version 0.11.0. |
|
|
Maximum number of partitions handled per writer. |
100 |
|
Target maximum size of written files; the actual size may be larger. |
|
|
Use randomized, unique table locations. |
|
|
Maximum duration to wait for completion of dynamic filters during split generation. |
|
|
Whether schema locations should be deleted when Trino can’t determine whether they contain external files. |
|
|
A decimal value in the range (0, 1] used as a minimum for weights assigned to each split. A low value may improve performance on tables with small files. A higher value may improve performance for queries with highly skewed aggregations or joins. |
0.05 |
|
Enables Статистики оптимизатора. The equivalent
catalog session property
is |
|
|
Enable projection pushdown |
|
|
Catalog to redirect to when a Hive table is referenced. |
|
|
Schema for creating materialized views storage tables. When this property
is not configured, storage tables are created in the same schema as the
materialized view definition. When the |
Empty |
|
Enable to allow user to call |
|
ORC format configuration#
The following properties are used to configure the read and write operations with ORC files performed by the Iceberg connector.
Property name |
Description |
Default |
|---|---|---|
|
Enable bloom filters for predicate pushdown. |
|
Parquet format configuration#
The following properties are used to configure the read and write operations with Parquet files performed by the Iceberg connector.
Property Name |
Description |
Default |
|---|---|---|
|
Sets the maximum number of rows read in a batch. |
|
|
Whether batched column readers should be used when reading Parquet files
for improved performance. Set this property to |
|
Authorization checks#
You can enable authorization checks for the connector by setting
the iceberg.security property in the catalog properties file. This
property must be one of the following values:
Property value |
Description |
|---|---|
|
No authorization checks are enforced. |
|
The connector relies on system-level access control. |
|
Operations that read data or metadata, such as SELECT are permitted. No operations that write data or metadata, such as CREATE TABLE, INSERT, or DELETE are allowed. |
|
Authorization checks are enforced using a catalog-level access control
configuration file whose path is specified in the |
Table redirection#
Trino offers the possibility to transparently redirect operations on an existing table to the appropriate catalog based on the format of the table and catalog configuration.
In the context of connectors which depend on a metastore service (for example, Hive коннектор, Iceberg коннектор and Delta Lake коннектор), the metastore (Hive metastore service, AWS Glue Data Catalog) can be used to accustom tables with different table formats. Therefore, a metastore database can hold a variety of tables with different table formats.
As a concrete example, let’s use the following simple scenario which makes use of table redirection:
USE example.example_schema;
EXPLAIN SELECT * FROM example_table;
Query Plan
-------------------------------------------------------------------------
Fragment 0 [SOURCE]
...
Output[columnNames = [...]]
│ ...
└─ TableScan[table = another_catalog:example_schema:example_table]
...
The output of the EXPLAIN statement points out the actual
catalog which is handling the SELECT query over the table example_table.
The table redirection functionality works also when using fully qualified names for the tables:
EXPLAIN SELECT * FROM example.example_schema.example_table;
Query Plan
-------------------------------------------------------------------------
Fragment 0 [SOURCE]
...
Output[columnNames = [...]]
│ ...
└─ TableScan[table = another_catalog:example_schema:example_table]
...
Trino offers table redirection support for the following operations:
Table read operations
Table write operations
Table management operations
Trino does not offer view redirection support.
The connector supports redirection from Iceberg tables to Hive tables
with the iceberg.hive-catalog-name catalog configuration property.
SQL support#
This connector provides read access and write access to data and metadata in Iceberg. In addition to the globally available and read operation statements, the connector supports the following features:
Schema and table management#
The Безопасность functionality includes support for:
CREATE SCHEMA#
The connector supports creating schemas. You can create a schema with or without a specified location.
You can create a schema with the CREATE SCHEMA statement and the
location schema property. The tables in this schema, which have no explicit
location set in CREATE TABLE statement, are located in a
subdirectory under the directory corresponding to the schema location.
Create a schema on S3:
CREATE SCHEMA example.example_s3_schema
WITH (location = 's3://my-bucket/a/path/');
Create a schema on a S3 compatible object storage such as MinIO:
CREATE SCHEMA example.example_s3a_schema
WITH (location = 's3a://my-bucket/a/path/');
Create a schema on HDFS:
CREATE SCHEMA example.example_hdfs_schema
WITH (location='hdfs://hadoop-master:9000/user/hive/warehouse/a/path/');
Optionally, on HDFS, the location can be omitted:
CREATE SCHEMA example.example_hdfs_schema;
Creating tables#
The Iceberg connector supports creating tables using the CREATE TABLE syntax. Optionally specify the table properties supported by this connector:
CREATE TABLE example_table (
c1 integer,
c2 date,
c3 double
)
WITH (
format = 'PARQUET',
partitioning = ARRAY['c1', 'c2'],
location = 's3://my-bucket/a/path/'
);
When the location table property is omitted, the content of the table
is stored in a subdirectory under the directory corresponding to the
schema location.
The Iceberg connector supports creating tables using the CREATE TABLE AS with SELECT syntax:
CREATE TABLE tiny_nation
WITH (
format = 'PARQUET'
)
AS
SELECT *
FROM nation
WHERE nationkey < 10;
Another flavor of creating tables with CREATE TABLE AS is with VALUES syntax:
CREATE TABLE yearly_clicks (
year,
clicks
)
WITH (
partitioning = ARRAY['year']
)
AS VALUES
(2021, 10000),
(2022, 20000);
NOT NULL column constraint#
The Iceberg connector supports setting NOT NULL constraints on the table columns.
The NOT NULL constraint can be set on the columns, while creating tables by
using the CREATE TABLE syntax:
CREATE TABLE example_table (
year INTEGER NOT NULL,
name VARCHAR NOT NULL,
age INTEGER,
address VARCHAR
);
When trying to insert/update data in the table, the query fails if trying
to set NULL value on a column having the NOT NULL constraint.
DROP TABLE#
The Iceberg connector supports dropping a table by using the DROP TABLE syntax. When the command succeeds, both the data of the Iceberg table and also the information related to the table in the metastore service are removed. Dropping tables which have their data/metadata stored in a different location than the table’s corresponding base directory on the object store is not supported.
ALTER TABLE EXECUTE#
The connector supports the following commands for use with ALTER TABLE EXECUTE.
optimize#
The optimize command is used for rewriting the active content
of the specified table so that it is merged into fewer but
larger files.
In case that the table is partitioned, the data compaction
acts separately on each partition selected for optimization.
This operation improves read performance.
All files with a size below the optional file_size_threshold
parameter (default value for the threshold is 100MB) are
merged:
ALTER TABLE test_table EXECUTE optimize
The following statement merges the files in a table that are under 10 megabytes in size:
ALTER TABLE test_table EXECUTE optimize(file_size_threshold => '10MB')
You can use a WHERE clause with the columns used to partition
the table, to apply optimize only on the partition(s) corresponding
to the filter:
ALTER TABLE test_partitioned_table EXECUTE optimize
WHERE partition_key = 1
expire_snapshots#
The expire_snapshots command removes all snapshots and all related metadata and data files.
Regularly expiring snapshots is recommended to delete data files that are no longer needed,
and to keep the size of table metadata small.
The procedure affects all snapshots that are older than the time period configured with the retention_threshold parameter.
expire_snapshots can be run as follows:
ALTER TABLE test_table EXECUTE expire_snapshots(retention_threshold => '7d')
The value for retention_threshold must be higher than or equal to iceberg.expire_snapshots.min-retention in the catalog
otherwise the procedure will fail with similar message:
Retention specified (1.00d) is shorter than the minimum retention configured in the system (7.00d).
The default value for this property is 7d.
remove_orphan_files#
The remove_orphan_files command removes all files from table’s data directory which are
not linked from metadata files and that are older than the value of retention_threshold parameter.
Deleting orphan files from time to time is recommended to keep size of table’s data directory under control.
remove_orphan_files can be run as follows:
ALTER TABLE test_table EXECUTE remove_orphan_files(retention_threshold => '7d')
The value for retention_threshold must be higher than or equal to iceberg.remove_orphan_files.min-retention in the catalog
otherwise the procedure will fail with similar message:
Retention specified (1.00d) is shorter than the minimum retention configured in the system (7.00d).
The default value for this property is 7d.
drop_extended_stats#
The drop_extended_stats command removes all extended statistics information from
the table.
drop_extended_stats can be run as follows:
ALTER TABLE test_table EXECUTE drop_extended_stats
ALTER TABLE SET PROPERTIES#
The connector supports modifying the properties on existing tables using ALTER TABLE SET PROPERTIES.
The following table properties can be updated after a table is created:
formatformat_versionpartitioning
For example, to update a table from v1 of the Iceberg specification to v2:
ALTER TABLE table_name SET PROPERTIES format_version = 2;
Or to set the column my_new_partition_column as a partition column on a table:
ALTER TABLE table_name SET PROPERTIES partitioning = ARRAY[<existing partition columns>, 'my_new_partition_column'];
The current values of a table’s properties can be shown using SHOW CREATE TABLE.
COMMENT#
The Iceberg connector supports setting comments on the following objects:
tables
views
table columns
The COMMENT option is supported on both the table and
the table columns for the CREATE TABLE operation.
The COMMENT option is supported for adding table columns
through the ALTER TABLE operations.
The connector supports the command COMMENT for setting comments on existing entities.
Data management#
The DML functionality includes support for INSERT,
UPDATE, DELETE, and MERGE statements.
Deletion by partition#
For partitioned tables, the Iceberg connector supports the deletion of entire
partitions if the WHERE clause specifies filters only on the identity-transformed
partitioning columns, that can match entire partitions. Given the table definition
from Partitioned Tables section,
the following SQL statement deletes all partitions for which country is US:
DELETE FROM example.testdb.customer_orders
WHERE country = 'US'
A partition delete is performed if the WHERE clause meets these conditions.
Row level deletion#
Tables using v2 of the Iceberg specification support deletion of individual rows by writing position delete files.
Type mapping#
The connector reads and writes data into the supported data file formats Avro, ORC, and Parquet, following the Iceberg specification.
Because Trino and Iceberg each support types that the other does not, this connector modifies some types when reading or writing data. Data types may not map the same way in both directions between Trino and the data source. Refer to the following sections for type mapping in each direction.
The Iceberg specification includes supported data types and the mapping to the formating in the Avro, ORC, or Parquet files:
Iceberg to Trino type mapping#
The connector maps Iceberg types to the corresponding Trino types following this table:
Iceberg type |
Trino type |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
No other types are supported.
Trino to Iceberg type mapping#
The connector maps Trino types to the corresponding Iceberg types following this table:
Trino type |
Iceberg type |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
No other types are supported.
Partitioned tables#
Iceberg supports partitioning by specifying transforms over the table columns. A partition is created for each unique tuple value produced by the transforms. Identity transforms are simply the column name. Other transforms are:
Transform |
Description |
|---|---|
|
A partition is created for each year. The partition value is the
integer difference in years between |
|
A partition is created for each month of each year. The partition
value is the integer difference in months between |
|
A partition is created for each day of each year. The partition
value is the integer difference in days between |
|
A partition is created hour of each day. The partition value is a timestamp with the minutes and seconds set to zero. |
|
The data is hashed into the specified number of buckets. The
partition value is an integer hash of |
|
The partition value is the first |
In this example, the table is partitioned by the month of order_date, a hash of
account_number (with 10 buckets), and country:
CREATE TABLE example.testdb.customer_orders (
order_id BIGINT,
order_date DATE,
account_number BIGINT,
customer VARCHAR,
country VARCHAR)
WITH (partitioning = ARRAY['month(order_date)', 'bucket(account_number, 10)', 'country'])
Rolling back to a previous snapshot#
Iceberg supports a «snapshot» model of data, where table snapshots are identified by a snapshot ID.
The connector provides a system table exposing snapshot information for every
Iceberg table. Snapshots are identified by BIGINT snapshot IDs.
For example, you could find the snapshot IDs for the customer_orders table
by running the following query:
SELECT snapshot_id
FROM example.testdb."customer_orders$snapshots"
ORDER BY committed_at DESC
Time travel queries#
The connector offers the ability to query historical data. This allows you to query the table as it was when a previous snapshot of the table was taken, even if the data has since been modified or deleted.
The historical data of the table can be retrieved by specifying the snapshot identifier corresponding to the version of the table that needs to be retrieved:
SELECT *
FROM example.testdb.customer_orders FOR VERSION AS OF 8954597067493422955
A different approach of retrieving historical data is to specify a point in time in the past, such as a day or week ago. The latest snapshot of the table taken before or at the specified timestamp in the query is internally used for providing the previous state of the table:
SELECT *
FROM example.testdb.customer_orders FOR TIMESTAMP AS OF TIMESTAMP '2022-03-23 09:59:29.803 Europe/Vienna'
Rolling back to a previous snapshot#
Use the $snapshots metadata table to determine the latest snapshot ID of the table like in the following query:
SELECT snapshot_id
FROM example.testdb."customer_orders$snapshots"
ORDER BY committed_at DESC LIMIT 1
A SQL procedure system.rollback_to_snapshot allows the caller to roll back
the state of the table to a previous snapshot id:
CALL example.system.rollback_to_snapshot('testdb', 'customer_orders', 8954597067493422955)
Schema evolution#
Iceberg supports schema evolution, with safe column add, drop, reorder and rename operations, including in nested structures. Table partitioning can also be changed and the connector can still query data created before the partitioning change.
Register table#
The connector can register existing Iceberg tables with the catalog.
The procedure system.register_table allows the caller to register an
existing Iceberg table in the metastore, using its existing metadata and data
files:
CALL example.system.register_table(schema_name => 'testdb', table_name => 'customer_orders', table_location => 'hdfs://hadoop-master:9000/user/hive/warehouse/customer_orders-581fad8517934af6be1857a903559d44')
In addition, you can provide a file name to register a table with specific metadata. This may be used to register the table with some specific table state, or may be necessary if the connector cannot automatically figure out the metadata version to use:
CALL example.system.register_table(schema_name => 'testdb', table_name => 'customer_orders', table_location => 'hdfs://hadoop-master:9000/user/hive/warehouse/customer_orders-581fad8517934af6be1857a903559d44', metadata_file_name => '00003-409702ba-4735-4645-8f14-09537cc0b2c8.metadata.json')
To prevent unauthorized users from accessing data, this procedure is disabled by default.
The procedure is enabled only when iceberg.register-table-procedure.enabled is set to true.
Migrating existing tables#
The connector can read from or write to Hive tables that have been migrated to Iceberg. There is no Trino support for migrating Hive tables to Iceberg, so you need to either use the Iceberg API or Apache Spark.
Iceberg table properties#
Property name |
Description |
|---|---|
|
Optionally specifies the format of table data files;
either |
|
Optionally specifies table partitioning.
If a table is partitioned by columns |
|
Optionally specifies the file system location URI for the table. |
|
Optionally specifies the format version of the Iceberg
specification to use for new tables; either |
|
Comma separated list of columns to use for ORC bloom filter.
It improves the performance of queries using Equality and IN predicates
when reading ORC file.
Requires ORC format.
Defaults to |
|
The ORC bloom filters false positive probability.
Requires ORC format.
Defaults to |
The table definition below specifies format Parquet, partitioning by columns c1 and c2,
and a file system location of /var/example_tables/test_table:
CREATE TABLE test_table (
c1 integer,
c2 date,
c3 double)
WITH (
format = 'PARQUET',
partitioning = ARRAY['c1', 'c2'],
location = '/var/example_tables/test_table')
The table definition below specifies format ORC, bloom filter index by columns c1 and c2,
fpp is 0.05, and a file system location of /var/example_tables/test_table:
CREATE TABLE test_table (
c1 integer,
c2 date,
c3 double)
WITH (
format = 'ORC',
location = '/var/example_tables/test_table',
orc_bloom_filter_columns = ARRAY['c1', 'c2'],
orc_bloom_filter_fpp = 0.05)
Metadata columns#
In addition to the defined columns, the Iceberg connector automatically exposes path metadata as a hidden column in each table:
$path: Full file system path name of the file for this row$file_modified_time: Timestamp of the last modification of the file for this row
You can use these columns in your SQL statements like any other column. This can be selected directly, or used in conditional statements. For example, you can inspect the file path for each record:
SELECT *, "$path", "$file_modified_time"
FROM example.web.page_views;
Retrieve all records that belong to a specific file using "$path" filter:
SELECT *
FROM example.web.page_views
WHERE "$path" = '/usr/iceberg/table/web.page_views/data/file_01.parquet'
Retrieve all records that belong to a specific file using "$file_modified_time" filter:
SELECT *
FROM example.web.page_views
WHERE "$file_modified_time" = CAST('2022-07-01 01:02:03.456 UTC' AS timestamp with time zone)
Metadata tables#
The connector exposes several metadata tables for each Iceberg table. These metadata tables contain information about the internal structure of the Iceberg table. You can query each metadata table by appending the metadata table name to the table name:
SELECT * FROM "test_table$data"
$data table#
The $data table is an alias for the Iceberg table itself.
The statement:
SELECT * FROM "test_table$data"
is equivalent to:
SELECT * FROM test_table
$properties table#
The $properties table provides access to general information about Iceberg
table configuration and any additional metadata key/value pairs that the table
is tagged with.
You can retrieve the properties of the current snapshot of the Iceberg
table test_table by using the following query:
SELECT * FROM "test_table$properties"
key | value |
-----------------------+----------+
write.format.default | PARQUET |
$history table#
The $history table provides a log of the metadata changes performed on
the Iceberg table.
You can retrieve the changelog of the Iceberg table test_table
by using the following query:
SELECT * FROM "test_table$history"
made_current_at | snapshot_id | parent_id | is_current_ancestor
----------------------------------+----------------------+----------------------+--------------------
2022-01-10 08:11:20 Europe/Vienna | 8667764846443717831 | <null> | true
2022-01-10 08:11:34 Europe/Vienna | 7860805980949777961 | 8667764846443717831 | true
The output of the query has the following columns:
Name |
Type |
Description |
|---|---|---|
|
|
The time when the snapshot became active |
|
|
The identifier of the snapshot |
|
|
The identifier of the parent snapshot |
|
|
Whether or not this snapshot is an ancestor of the current snapshot |
$snapshots table#
The $snapshots table provides a detailed view of snapshots of the
Iceberg table. A snapshot consists of one or more file manifests,
and the complete table contents is represented by the union
of all the data files in those manifests.
You can retrieve the information about the snapshots of the Iceberg table
test_table by using the following query:
SELECT * FROM "test_table$snapshots"
committed_at | snapshot_id | parent_id | operation | manifest_list | summary
----------------------------------+----------------------+----------------------+--------------------+------------------------------------------------------------------------------------------------------------------------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
2022-01-10 08:11:20 Europe/Vienna | 8667764846443717831 | <null> | append | hdfs://hadoop-master:9000/user/hive/warehouse/test_table/metadata/snap-8667764846443717831-1-100cf97e-6d56-446e-8961-afdaded63bc4.avro | {changed-partition-count=0, total-equality-deletes=0, total-position-deletes=0, total-delete-files=0, total-files-size=0, total-records=0, total-data-files=0}
2022-01-10 08:11:34 Europe/Vienna | 7860805980949777961 | 8667764846443717831 | append | hdfs://hadoop-master:9000/user/hive/warehouse/test_table/metadata/snap-7860805980949777961-1-faa19903-1455-4bb8-855a-61a1bbafbaa7.avro | {changed-partition-count=1, added-data-files=1, total-equality-deletes=0, added-records=1, total-position-deletes=0, added-files-size=442, total-delete-files=0, total-files-size=442, total-records=1, total-data-files=1}
The output of the query has the following columns:
Name |
Type |
Description |
|---|---|---|
|
|
The time when the snapshot became active |
|
|
The identifier for the snapshot |
|
|
The identifier for the parent snapshot |
|
|
The type of operation performed on the Iceberg table. The supported operation types in Iceberg are:
|
|
|
The list of avro manifest files containing the detailed information about the snapshot changes. |
|
|
A summary of the changes made from the previous snapshot to the current snapshot |
$manifests table#
The $manifests table provides a detailed overview of the manifests
corresponding to the snapshots performed in the log of the Iceberg table.
You can retrieve the information about the manifests of the Iceberg table
test_table by using the following query:
SELECT * FROM "test_table$manifests"
path | length | partition_spec_id | added_snapshot_id | added_data_files_count | added_rows_count | existing_data_files_count | existing_rows_count | deleted_data_files_count | deleted_rows_count | partitions
----------------------------------------------------------------------------------------------------------------+-----------------+----------------------+-----------------------+-------------------------+------------------+-----------------------------+---------------------+-----------------------------+--------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------
hdfs://hadoop-master:9000/user/hive/warehouse/test_table/metadata/faa19903-1455-4bb8-855a-61a1bbafbaa7-m0.avro | 6277 | 0 | 7860805980949777961 | 1 | 100 | 0 | 0 | 0 | 0 | {{contains_null=false, contains_nan= false, lower_bound=1, upper_bound=1},{contains_null=false, contains_nan= false, lower_bound=2021-01-12, upper_bound=2021-01-12}}
The output of the query has the following columns:
Name |
Type |
Description |
|---|---|---|
|
|
The manifest file location |
|
|
The manifest file length |
|
|
The identifier for the partition specification used to write the manifest file |
|
|
The identifier of the snapshot during which this manifest entry has been added |
|
|
The number of data files with status |
|
|
The total number of rows in all data files with status |
|
|
The number of data files with status |
|
|
The total number of rows in all data files with status |
|
|
The number of data files with status |
|
|
The total number of rows in all data files with status |
|
|
Partition range metadata |
$partitions table#
The $partitions table provides a detailed overview of the partitions
of the Iceberg table.
You can retrieve the information about the partitions of the Iceberg table
test_table by using the following query:
SELECT * FROM "test_table$partitions"
partition | record_count | file_count | total_size | data
-----------------------+---------------+---------------+---------------+------------------------------------------------------
{c1=1, c2=2021-01-12} | 2 | 2 | 884 | {c3={min=1.0, max=2.0, null_count=0, nan_count=NULL}}
{c1=1, c2=2021-01-13} | 1 | 1 | 442 | {c3={min=1.0, max=1.0, null_count=0, nan_count=NULL}}
The output of the query has the following columns:
Name |
Type |
Description |
|---|---|---|
|
|
A row which contains the mapping of the partition column name(s) to the partition column value(s) |
|
|
The number of records in the partition |
|
|
The number of files mapped in the partition |
|
|
The size of all the files in the partition |
|
|
Partition range metadata |
$files table#
The $files table provides a detailed overview of the data files in current snapshot of the Iceberg table.
To retrieve the information about the data files of the Iceberg table test_table use the following query:
SELECT * FROM "test_table$files"
content | file_path | record_count | file_format | file_size_in_bytes | column_sizes | value_counts | null_value_counts | nan_value_counts | lower_bounds | upper_bounds | key_metadata | split_offsets | equality_ids
----------+-------------------------------------------------------------------------------------------------------------------------------+-----------------+---------------+----------------------+----------------------+-------------------+--------------------+-------------------+-----------------------------+-----------------------------+----------------+----------------+---------------
0 | hdfs://hadoop-master:9000/user/hive/warehouse/test_table/data/c1=3/c2=2021-01-14/af9872b2-40f3-428f-9c87-186d2750d84e.parquet | 1 | PARQUET | 442 | {1=40, 2=40, 3=44} | {1=1, 2=1, 3=1} | {1=0, 2=0, 3=0} | <null> | {1=3, 2=2021-01-14, 3=1.3} | {1=3, 2=2021-01-14, 3=1.3} | <null> | <null> | <null>
The output of the query has the following columns:
Name |
Type |
Description |
|---|---|---|
|
|
Type of content stored in the file. The supported content types in Iceberg are:
|
|
|
The data file location |
|
|
The format of the data file |
|
|
The number of entries contained in the data file |
|
|
The data file size |
|
|
Mapping between the Iceberg column ID and its corresponding size in the file |
|
|
Mapping between the Iceberg column ID and its corresponding count of entries in the file |
|
|
Mapping between the Iceberg column ID and its corresponding count of |
|
|
Mapping between the Iceberg column ID and its corresponding count of non numerical values in the file |
|
|
Mapping between the Iceberg column ID and its corresponding lower bound in the file |
|
|
Mapping between the Iceberg column ID and its corresponding upper bound in the file |
|
|
Metadata about the encryption key used to encrypt this file, if applicable |
|
|
List of recommended split locations |
|
|
The set of field IDs used for equality comparison in equality delete files |
Materialized views#
The Iceberg connector supports Материализованные представления (materialized view). In the underlying system each materialized view consists of a view definition and an Iceberg storage table. The storage table name is stored as a materialized view property. The data is stored in that storage table.
You can use the Iceberg table properties to control the created storage
table and therefore the layout and performance. For example, you can use the
following clause with CREATE MATERIALIZED VIEW to use the ORC format
for the data files and partition the storage per day using the column
_date:
WITH ( format = 'ORC', partitioning = ARRAY['event_date'] )
By default, the storage table is created in the same schema as the materialized
view definition. The iceberg.materialized-views.storage-schema catalog
configuration property or storage_schema materialized view property can be
used to specify the schema where the storage table will be created.
Updating the data in the materialized view with REFRESH MATERIALIZED VIEW deletes the data from the storage table, and inserts the data that is the result of executing the materialized view query into the existing table. Data is replaced atomically, so users can continue to query the materialized view while it is being refreshed. Refreshing a materialized view also stores the snapshot-ids of all Iceberg tables that are part of the materialized view’s query in the materialized view metadata. When the materialized view is queried, the snapshot-ids are used to check if the data in the storage table is up to date. If the data is outdated, the materialized view behaves like a normal view, and the data is queried directly from the base tables. Detecting outdated data is possible only when the materialized view uses Iceberg tables only, or when it uses mix of Iceberg and non-Iceberg tables but some Iceberg tables are outdated. When the materialized view is based on non-Iceberg tables, querying it can return outdated data, since the connector has no information whether the underlying non-Iceberg tables have changed.
Dropping a materialized view with DROP MATERIALIZED VIEW removes the definition and the storage table.
Table statistics#
The Iceberg connector can collect column statistics using ANALYZE
statement. This can be disabled using iceberg.extended-statistics.enabled
catalog configuration property, or the corresponding
extended_statistics_enabled session property.
Updating table statistics#
If your queries are complex and include joining large data sets, running ANALYZE on tables may improve query performance by collecting statistical information about the data:
ANALYZE table_name
This query collects statistics for all columns.
On wide tables, collecting statistics for all columns can be expensive.
It is also typically unnecessary - statistics are
only useful on specific columns, like join keys, predicates, or grouping keys. You can
specify a subset of columns to analyzed with the optional columns property:
ANALYZE table_name WITH (columns = ARRAY['col_1', 'col_2'])
This query collects statistics for columns col_1 and col_2.
Note that if statistics were previously collected for all columns, they need to be dropped using drop_extended_stats command before re-analyzing.