The Cobalt database

Every quest runs against a real SQL Server database seeded with the defects from the story: duplicate loads, split payments, timezone drift, NULLs, and orphaned legacy rows.

Setup

1. Install SQL Server Developer Edition, Express, or use the Azure SQL free tier.
2. Open SSMS, Azure Data Studio, or VS Code with the MSSQL extension.
3. Run the scripts in order: 01_schema.sql, 02_seed.sql, 03_grading-harness.sql.
4. Run the verification queries below. If the counts match, you are ready for Day 1.

Verification

USE CobaltFiles;

SELECT COUNT(*) AS Orders        FROM sales.Orders;        -- 4,200
SELECT COUNT(*) AS Customers     FROM core.Customers;     -- 425 (400 people + 25 duplicates)
SELECT COUNT(*) AS Payments      FROM sales.Payments;     -- 5,040
SELECT COUNT(*) AS LegacyRows    FROM staging.LegacyOrderDump; -- 500
SELECT * FROM reporting.LegacyRevenue;                    -- the report that starts the mystery

Expected row counts