Cloudera

FEBRUARY 14, 2024

This blog discusses a few problems that you might encounter with Iceberg tables and offers strategies on how to optimize them in each of those scenarios. Problem with too many snapshots Everytime a write operation occurs on an Iceberg table, a new snapshot is created. See Write properties.

Strategy 57

Strategy Optimization Snapshot Metadata 57

Cloudera

FEBRUARY 8, 2024

Queries containing joins, filters, projections, group-by, or aggregations without group-by can be transparently rewritten by the Hive optimizer to use one or more eligible materialized views. The snapshotId of the source tables involved in the materialized view are also maintained in the metadata.

Snapshot 90

Snapshot Metadata Cost-Benefit Data Warehouse 90

Cloudera

APRIL 3, 2023

Every table change creates an Iceberg snapshot, this helps to resolve concurrency issues and allows readers to scan a stable table state every time. The table metadata is stored next to the data files under a metadata directory, which allows multiple engines to use the same table simultaneously. ID, TBL_ICEBERG_PART_2.NAME,

Data Warehouse 81

Data Warehouse Snapshot Metadata Cost-Benefit 81

Cloudera

JULY 13, 2023

A range of Iceberg table analysis such as listing table’s data file, selecting table snapshot, partition filtering, and predicate filtering can be delegated through Iceberg Java API instead, obviating the need for each query engine to implement it themself. The data files and metadata files in Iceberg format are immutable.

Metadata 70

Metadata Snapshot Data Warehouse Statistics 70

Cloudera

JULY 14, 2023

They also provide a “ snapshot” procedure that creates an Iceberg table with a different name with the same underlying data. You could first create a snapshot table, run sanity checks on the snapshot table, and ensure that everything is in order. Hive creates Iceberg’s metadata files for the same exact table.

Snapshot 70

Snapshot Metadata Data Warehouse Testing 70