Learner track
The hands-on modules you work through, from an empty laptop to a migrated and benchmarked workload.
Work the modules in order — each one names a starting checkpoint, so you always know what needs to be true before you begin it. By the end you will have profiled a Snowflake workload, made explicit engine and schema decisions, executed the migration, rebuilt the dbt pipeline on ClickHouse, benchmarked the result, and cut a live producer over from one cloud to the other.
Keep the Snowflake producer running through module 05
Module 01 starts a producer that writes live trip rows into Snowflake. It must keep running through module 05 — do not stop it, and do not let it stop, anywhere in modules 02, 03, or 04. Module 05's cutover step measures the exact gap between when writes stop going to Snowflake and when they resume on ClickHouse. Stop the producer early and there is nothing for that step to measure — this is the single most likely way to break the lab.
Modules
- Setup — install the toolchain, create the two cloud trial accounts, clone the repo, and build both dbt virtualenvs.
- Source environment — provision a Snowflake environment that mirrors a real customer deployment, and start the live trip producer.
- Plan and design — profile the Snowflake workload and make the architecture decisions the migration will execute.
- Provision and migrate — provision ClickHouse Cloud and move 50 million rows with a resumable migration script.
- Rebuild the dbt pipeline — rebuild the Medallion pipeline on ClickHouse with dbt-clickhouse.
- Benchmark and cutover — benchmark both engines, cut the producer over, verify parity, and tear down.
- Evaluation — the open-book assessment that earns the ClickHouse Migration Proficiency Badge.
If something breaks
Troubleshooting is a symptom, cause, and fix reference for the failures learners actually hit while running this lab, grouped by the module where they surface. Check there before asking for help — the fix is likely already written down.