The Hidden Costs and Concerns of Iceberg Maintenance

Everyone talks about how great Apache Iceberg is, but nobody warns you about this: without proper maintenance, your tables will bloat, queries will slow down, and your catalog will run out of memory. Here are the 4 critical operations you MUST run regularly.

Expiring snapshots prevents metadata bloat (Datadog learned this the hard way with catalog memory pressure). Deleting orphan files cleans up failed writes. Compacting data files keeps streaming workloads fast. Compacting manifests optimizes query planning.

The hard part? Knowing when to run each operation. Too aggressive and you waste compute. Too passive and you tank performance. This clip breaks down the real operational requirements that production Iceberg deployments demand.