06 Evaluation

Facilitator companion to the learner lesson 06 Evaluation.

Timing

About 60 minutes (the lab's own estimate is 45-60 minutes for the open-book assessment itself). Nothing here is unattended in the scripted sense — it is a partner writing answers — so there is no scripted pause to plan a break around. Facilitators are typically not present for the writing itself; the 60 minutes is the partner's own time, not seat time in front of the room.

Talk track

• This module is not credit for having run the scripts — modules 01-05 already proved the migration mechanically works. It certifies the partner can explain and defend the decisions behind it: profiling the workload, making the engine and schema calls, executing the migration, validating the result.
• It is open-book on purpose: migration-plan.md, the benchmark CSV, and every module of this playbook are all fair game. The open questions specifically reward depth of reasoning over recall — a partner who reaches a different but defensible conclusion should score well.
• State the scoring plainly so a partner knows what the 60 minutes buys them: 20 multiple-choice questions at 4 points each (80 points total), 4 open questions at 5 points each (20 points total), 100 points total, 80 to pass.
• The answer key lives at workshops/snowflake_migration_lab/04-evaluation/docs/answer-key.md. It is gitignored and deliberately not published to this site — it is distributed to Solutions Architects out of band, not through this playbook. Never point a partner at that path. The key is also not part of the public workshop_public/snowflake_migration_lab lab repository — the scripts, Terraform, and dbt projects that repository does carry were extracted separately from it, and it did not travel with them. It remains distributed to Solutions Architects out of band, exactly as above.

Common failures

• A partner tore down both cloud environments in module 05 without saving migration-plan.md and the benchmark CSV first. Both files are the only two things this module needs, and both are gone once module 05's teardown has run — there is no cloud environment left to regenerate them from. The only fix is to re-run modules 02 through 05 far enough to reproduce them, which is not a 60-minute problem.
• A partner edits assessment.md directly instead of copying it to assessment_<their name>.md first. The learner instructions call this out for a reason: only the copy is meant to hold answers, and the blank template is what the next partner, or the next cohort, still needs to find in the repo.
• TODO: this module's own README has no Troubleshooting section. Capture any grading or submission friction (missing attachments, ambiguous open-answer wording, partners asking for hints) once this has been run with a live cohort.

Reset steps

• If a partner scores 70-79 (near-miss): the Solutions Architect gives targeted feedback and the partner gets one retry, limited to the open questions.
• If a partner scores below 70: the Solutions Architect identifies weak areas and the partner retakes the full assessment after review.
• If assessment.md was edited in place instead of copied, restore the blank template from the repository before the next partner needs it — do not hand out a partially-answered file as the "blank" one.
• There is no cloud environment to reset for this module — module 05's teardown already ran. If a partner needs Part 1-3 artifacts they did not save, the recovery is re-running the relevant earlier module, not anything in this one.