Target Snowflake

Table Naming Convention

{database}.{schema}.{table}

Example: MIGRATION_DB.BRONZE.CUSTOMERS

Medallion Architecture

RAW     -> Raw ingestion (staging)
BRONZE  -> Cleansed, typed
SILVER  -> Conformed, business rules
GOLD    -> Aggregated, dimensional

Planned Code Patterns

Snowpark Python Pattern

from snowflake.snowpark import Session
from snowflake.snowpark.functions import col, lit, when

# Connection
connection_parameters = {
    "account": "xy12345.us-east-1",
    "user": "migration_user",
    "password": "...",
    "warehouse": "COMPUTE_WH",
    "database": "MIGRATION_DB",
    "schema": "BRONZE",
    "role": "TRANSFORMER"
}

session = Session.builder.configs(connection_parameters).create()

# Read
df = session.table("source_table")

# Transform
df_transformed = df.with_column("new_col", col("old_col") * 2)

# Write
df_transformed.write.mode("overwrite").save_as_table("target_table")

SQL Pattern

-- Merge pattern (upsert)
MERGE INTO target_table t
USING source_table s
ON t.id = s.id
WHEN MATCHED THEN UPDATE SET t.col1 = s.col1
WHEN NOT MATCHED THEN INSERT (id, col1) VALUES (s.id, s.col1);

Planned Type Mappings

Environment Variables (Planned)

Contributing

To implement this skill:

1. Create code-patterns.md with Snowpark and SQL templates
2. Create type-mappings.json for source systems
3. Create execution scripts (submit_query.py, etc.)
4. Test with sample migrations