Iceberg Table Maintenance Calculator

Describe one Apache Iceberg table and this page works out how often to compact it, what file size to aim for, how long to keep snapshots, how often to clean up orphan files, and writes the Spark SQL CALL statements with those arguments filled in. It runs entirely in your browser; nothing you type leaves the page.

Your table

Table name (catalog.namespace.table)Daily ingest volume (GB per day, compressed on storage)Commits per dayPartitions written per commit

Targets and retention

Time travel window you need (days)Snapshots always kept for rollbackLongest write or compaction job (hours)

How to read the output

The numbers come from two kinds of input, and the page keeps them apart:

• Documented Iceberg behavior. Procedure names, argument names, option names and defaults are the ones in the Iceberg 1.11.0 documentation: the 512 MB default target file size, the 75% small-file threshold, min-input-files of 5, snapshot expiry defaults of 5 days and 1 snapshot, the 3-day orphan window, and the 100-manifest merge threshold.
• Rules of thumb. The cadence thresholds (at most about 50 small files per partition between compaction runs, the 1 GB and 10 GB per-partition-per-day steps for target file size, weekly versus monthly orphan and manifest runs) are starting points from running these tables, not Iceberg defaults. Watch the files and snapshots metadata tables for a week or two and adjust.

The file-size estimate assumes the best case of one file per partition per commit. Real writers with many tasks produce more, smaller files, so if anything your table needs compaction more often than shown, not less.

The statements use the Spark procedure syntax CALL <catalog>.system.<procedure>(...) with named arguments. On Spark 4, procedure names are case-sensitive, so keep system and the procedure names in lower case.

Walk through each procedure

These pages run each step by hand against a small local table, with checks you can use to confirm it worked:

• Compact data files with rewrite_data_files
• Expire snapshots with expire_snapshots
• Create an Iceberg table to practice on

Sources

Checked against the Iceberg 1.11.0 documentation on September 29, 2026:

• Spark procedures: rewrite_data_files options, expire_snapshots, remove_orphan_files, rewrite_manifests
• Maintenance: the orphan file retention warning and metadata file cleanup
• Configuration: write.target-file-size-bytes, history.expire.*, write.metadata.*, commit.manifest.min-count-to-merge