Apache Iceberg compatibility

This page compares Query Engine with the Apache Iceberg specification and the features listed on Iceberg’s implementation status page. Read means Query Engine can query tables that use the feature. Write means Query Engine can create or change them. SQL commands and procedures follow the syntax in Iceberg’s Spark docs. To set up an Iceberg catalog, see Table formats.

Yes Supported Partial Supported with limits No Not supported yet

Summary

Area	Yes	Partial	No
Table spec features	18	1	71
Data types	14	0	32
DDL	0	0	37
DML	0	0	16
Queries and metadata tables	19	0	3
Procedures and maintenance	0	0	21
Catalog implementations	3	1	12
REST catalog operations	8	0	17
File IO	4	0	1

Table spec features

Rows follow the Apache Iceberg table spec and concept docs. Spec = format version that introduced it; — = not tied to one version.

Format versions

Feature	Iceberg spec	e6data read	e6data write
Format version 1	V1	Yes	No
Format version 2 (row-level deletes)	V2	Partial	No
Format version 3 (extended types and capabilities)	V3	No	No

File formats

Feature	Iceberg spec	e6data read	e6data write
Parquet data files	V1	Yes	No
ORC data files	V1	No	No
Avro data files	V1	No	No
Puffin files (statistics, deletion vectors)	—	No	No

Schemas and evolution

Feature	Iceberg spec	e6data read	e6data write
Add column (incl. nested fields)	V1	Yes	No
Drop column	V1	Yes	No
Rename column	V1	Yes	No
Reorder columns	V1	Yes	No
Type promotion: int → long	V1	No	No
Type promotion: float → double	V1	No	No
Type promotion: widen decimal precision	V1	No	No
Required → optional	V1	Yes	No
Identifier fields	—	No	No
Name mapping for files without field IDs	—	No	No
Column initial-default	V3	No	No
Column write-default	V3	No	No

Partitioning

Feature	Iceberg spec	e6data read	e6data write
Hidden partitioning (partition pruning from filters)	V1	Yes	No
Transform: identity	V1	No	No
Transform: bucket[N]	V1	No	No
Transform: truncate[W]	V1	No	No
Transform: year	V1	Yes	No
Transform: month	V1	Yes	No
Transform: day	V1	Yes	No
Transform: hour	V1	Yes	No
Transform: void	V1	No	No
Multi-argument transforms (partitioning and sorting)	V3	No	No
Partition evolution	V1	Yes	No

Sort orders

Feature	Iceberg spec	e6data read	e6data write
Table sort order	V1	Yes	No
Sort order evolution	V1	Yes	No

Row-level changes and lineage

Feature	Iceberg spec	e6data read	e6data write
Position delete files	V2	Yes	No
Equality delete files	V2	No	No
Binary deletion vectors	V3	No	No
Row lineage (_row_id, _last_updated_sequence_number)	V3	No	No

Snapshots and references

Feature	Iceberg spec	e6data read	e6data write
Snapshots and time travel	V1	No	No
Branches	—	No	No
Tags	—	No	No
Reference retention (max ref age, snapshots to keep)	—	No	No

Statistics and encryption

Feature	Iceberg spec	e6data read	e6data write
Column metrics in manifests (min / max / null counts)	V1	Yes	No
Table statistics files (Puffin, e.g. NDV sketches)	—	No	No
Partition statistics files	—	Yes	No
Table encryption keys	V3	No	No
AES GCM stream encryption	—	No	No

Data types

Full type list from the Iceberg spec and Implementation status page.

Primitive

Iceberg type	Iceberg spec	e6data read	e6data write
boolean	V1	Yes	No
int	V1	Yes	No
long	V1	Yes	No
float	V1	Yes	No
double	V1	Yes	No
decimal(P,S)	V1	Yes	No
string	V1	Yes	No
uuid	V1	No	No
fixed(L)	V1	Yes	No
binary	V1	No	No

Date and time

Iceberg type	Iceberg spec	e6data read	e6data write
date	V1	Yes	No
time	V1	No	No
timestamp	V1	Yes	No
timestamptz	V1	Yes	No
timestamp_ns	V3	No	No
timestamptz_ns	V3	No	No

Nested and semi-structured

Iceberg type	Iceberg spec	e6data read	e6data write
struct	V1	Yes	No
list	V1	Yes	No
map	V1	Yes	No
variant	V3	No	No

Geospatial and other

Iceberg type	Iceberg spec	e6data read	e6data write
geometry	V3	No	No
geography	V3	No	No
unknown	V3	No	No

DDL

Commands as documented in Iceberg’s Spark DDL docs. If e6data supports the capability with different syntax, mark Yes and note the syntax.

Create, replace and drop

Command	Iceberg spec	e6data
CREATE TABLE … USING iceberg	—	No
PARTITIONED BY with transforms	—	No
LOCATION / COMMENT / TBLPROPERTIES	—	No
CREATE TABLE AS SELECT (CTAS)	—	No
REPLACE TABLE AS SELECT (RTAS)	—	No
CREATE TABLE … LIKE	—	No
DROP TABLE	—	No
DROP TABLE … PURGE	—	No

ALTER TABLE: table and columns

Command	Iceberg spec	e6data
RENAME TO	—	No
SET / UNSET TBLPROPERTIES	—	No
ADD COLUMN(S), incl. nested	—	No
RENAME COLUMN	—	No
ALTER COLUMN … TYPE (widening)	—	No
ALTER COLUMN … COMMENT	—	No
ALTER COLUMN … FIRST / AFTER	—	No
ALTER COLUMN … DROP NOT NULL	—	No
DROP COLUMN	—	No

ALTER TABLE: partitioning, ordering, identifiers

Command	Iceberg spec	e6data
ADD PARTITION FIELD	—	No
DROP PARTITION FIELD	—	No
REPLACE PARTITION FIELD	—	No
WRITE ORDERED BY	—	No
WRITE LOCALLY ORDERED BY	—	No
WRITE DISTRIBUTED BY PARTITION	—	No
WRITE UNORDERED	—	No
SET IDENTIFIER FIELDS	—	No
DROP IDENTIFIER FIELDS	—	No

Branches and tags

Command	Iceberg spec	e6data
CREATE [OR REPLACE] BRANCH	—	No
REPLACE BRANCH	—	No
DROP BRANCH	—	No
CREATE [OR REPLACE] TAG	—	No
DROP TAG	—	No

Views (Iceberg view spec)

Command	Iceberg spec	e6data
CREATE VIEW	—	No
CREATE OR REPLACE VIEW	—	No
ALTER VIEW … SET / UNSET TBLPROPERTIES	—	No
DROP VIEW	—	No
SHOW VIEWS	—	No
SHOW CREATE TABLE on a view	—	No

DML

Writes

Operation	Iceberg spec	e6data
INSERT INTO	—	No
INSERT OVERWRITE (dynamic)	—	No
INSERT OVERWRITE (static partitions)	—	No
MERGE INTO	—	No
↳ Copy-on-write	—	No
↳ Merge-on-read	—	No
DELETE FROM	—	No
↳ Copy-on-write	—	No
↳ Merge-on-read	—	No
UPDATE	—	No
↳ Copy-on-write	—	No
↳ Merge-on-read	—	No

Branch writes and concurrency

Operation	Iceberg spec	e6data
Write to a branch	—	No
Write-audit-publish (WAP)	—	No
Optimistic concurrency with retries	—	No
Serializable isolation	—	No

Queries and metadata tables

Reads and time travel

Query	Iceberg spec	e6data
SELECT from Iceberg table	—	Yes
VERSION AS OF <snapshot-id>	—	Yes
VERSION AS OF <branch or tag>	—	No
TIMESTAMP AS OF	—	Yes
Incremental read between snapshots	—	No
Time travel on metadata tables	—	Yes

Metadata tables

Query	Iceberg spec	e6data
history	—	Yes
metadata_log_entries	—	Yes
snapshots	—	Yes
entries	—	Yes
files	—	Yes
data_files	—	Yes
delete_files	—	Yes
manifests	—	Yes
partitions	—	Yes
position_deletes	—	No
refs	—	Yes
all_data_files	—	Yes
all_delete_files	—	Yes
all_files	—	Yes
all_entries	—	Yes
all_manifests	—	Yes

Procedures and maintenance

Procedures as documented in Iceberg’s Spark procedures docs (CALL catalog.system.<name>). An equivalent command in e6data counts as Yes; note the syntax.

Snapshot management

Procedure	Iceberg spec	e6data
rollback_to_snapshot	—	No
rollback_to_timestamp	—	No
set_current_snapshot	—	No
cherrypick_snapshot	—	No
publish_changes	—	No
fast_forward	—	No

Metadata management

Procedure	Iceberg spec	e6data
expire_snapshots	—	No
remove_orphan_files	—	No
rewrite_data_files (compaction)	—	No
↳ sort / z-order strategy	—	No
rewrite_manifests	—	No
rewrite_position_delete_files	—	No

Table migration

Procedure	Iceberg spec	e6data
snapshot	—	No
migrate	—	No
add_files	—	No
register_table	—	No

Metadata, CDC, statistics, replication

Procedure	Iceberg spec	e6data
ancestors_of	—	No
create_changelog_view (CDC)	—	No
compute_table_stats	—	No
compute_partition_stats	—	No
rewrite_table_path (replication)	—	No

Catalog implementations

Catalogs documented by Apache Iceberg. Third-party catalogs that speak the REST spec (e.g. Polaris, Gravitino, Lakekeeper) count under REST.

Catalog	Iceberg spec	e6data read	e6data write
REST catalog	—	Yes	Partial
Hive Metastore catalog	—	Yes	No
AWS Glue catalog	—	Yes	No
JDBC / SQL catalog	—	No	No
Hadoop (file-system) catalog	—	No	No
Nessie catalog	—	No	No
AWS DynamoDB catalog	—	No	No
Custom catalog	—	No	No

REST catalog operations

Operations listed on the Iceberg Implementation status page and REST Catalog spec.

Namespace operations

Operation	Iceberg spec	e6data
listNamespaces	—	Yes
createNamespace	—	No
dropNamespace	—	No
namespaceExists	—	No
updateNamespaceProperties	—	No
loadNamespaceMetadata	—	Yes

Table operations

Operation	Iceberg spec	e6data
listTables	—	Yes
createTable	—	No
dropTable	—	No
loadTable	—	Yes
updateTable (commit)	—	No
renameTable	—	Yes
tableExists	—	No
registerTable	—	No

View operations

Operation	Iceberg spec	e6data
createView	—	Yes
dropView	—	Yes
listViews	—	Yes
viewExists	—	No
replaceView	—	No
renameView	—	No

Other REST capabilities

Operation	Iceberg spec	e6data
Multi-table transaction commit	—	No
Credential vending (loadCredentials)	—	No
Server-side scan planning	—	No
Metrics reporting	—	No
OAuth2 authentication	—	No

File IO

Storage backends from the Implementation status page.

Storage	Iceberg spec	e6data
Local filesystem	—	Yes
Hadoop filesystem (HDFS)	—	No
S3 compatible	—	Yes
GCS compatible	—	Yes
ADLS compatible	—	Yes

​Summary

​Table spec features

​Format versions

​File formats

​Schemas and evolution

​Partitioning

​Sort orders

​Row-level changes and lineage

​Snapshots and references

​Statistics and encryption

​Data types

​Primitive

​Date and time

​Nested and semi-structured

​Geospatial and other

​DDL

​Create, replace and drop

​ALTER TABLE: table and columns

​ALTER TABLE: partitioning, ordering, identifiers

​Branches and tags

​Views (Iceberg view spec)

​DML

​Writes

​Branch writes and concurrency

​Queries and metadata tables

​Reads and time travel

​Metadata tables

​Procedures and maintenance

​Snapshot management

​Metadata management

​Table migration

​Metadata, CDC, statistics, replication

​Catalog implementations

​REST catalog operations

​Namespace operations

​Table operations

​View operations

​Other REST capabilities

​File IO