A hands-on walkthrough of building and operating an Apache Iceberg lakehouse on Dremio Cloud, part 14 of a 15-part masterclass series. Covers creating Iceberg tables with hidden partitioning, ingesting data via COPY INTO and INSERT SELECT (including federation from Postgres, MySQL, Oracle, MongoDB), and Dremio's Columnar Cloud Cache for accelerating repeated queries. Also demonstrates the semantic layer for governed views, Reflections for query acceleration, column/row-level governance, table optimization (compaction, snapshot expiry), time travel queries, metadata inspection, and Dremio's built-in AI agent plus MCP Server for natural-language-to-SQL analytics.
Table of contents
Table of ContentsGetting StartedThe Dremio PlatformEssential SQL OperationsAI-Powered AnalyticsTry Dremio Cloud free for 30 daysQuestions this post answers
How does Dremio's Columnar Cloud Cache (C3) speed up queries on Iceberg data in object storage?
C3 stores frequently accessed Iceberg columns on local NVMe SSDs attached to query engine nodes, so repeated queries read from local disk instead of remote object storage. This reduces latency from hundreds of milliseconds down to single-digit milliseconds. It works transparently, automatically tracking access patterns and caching the most-queried data without manual configuration. Teams tuning lakehouse query latency can track caching and performance techniques like this on daily.dev.
How do Dremio Reflections accelerate queries on Iceberg tables without changing the SQL?
Reflections precompute and store query results as optimized Iceberg tables on fast storage, and the query optimizer transparently substitutes them when a query matches the reflection's definition. For example, an aggregate reflection defined with dimensions and measures can turn a 30-second scan against raw data into a sub-1-second query, with no changes needed to the original SQL or application code. Developers evaluating query acceleration strategies can follow lakehouse tooling updates on daily.dev.
How can I query PostgreSQL and Apache Iceberg tables together without migrating all the data first?
Dremio's query federation lets you join Iceberg tables directly with tables in PostgreSQL, MySQL, Oracle, MongoDB, and S3 files in a single SQL query, without moving data into Iceberg first. This is useful during migrations: you can query legacy systems and Iceberg tables side by side and swap the underlying source incrementally once ready. Anyone planning an incremental lakehouse migration can track federation approaches like this on daily.dev.