OBT Design Optimizer

设计并优化“一张大表”（OBT）模式。

SKILL.md

--- frontmatter

name: OBT Design Optimizer
description: Designs and optimizes One Big Table (OBT) patterns
version: 1.0.0
category: Data Modeling
skillId: SK-DEA-020
allowed-tools:
  - Read
  - Write
  - Edit
  - Glob
  - Grep
  - Bash

OBT Design Optimizer

Overview

Designs and optimizes One Big Table (OBT) patterns. This skill balances denormalization benefits with maintainability for analytical use cases.

Capabilities

•Column selection optimization
•Denormalization strategy
•Nested/repeated field design (BigQuery)
•Clustering key selection
•Partition strategy
•Update frequency optimization
•Query pattern analysis
•Storage vs. performance tradeoffs

Input Schema

json

{
  "sourceModels": ["object"],
  "queryPatterns": ["object"],
  "platform": "snowflake|bigquery|redshift",
  "constraints": {
    "maxColumns": "number",
    "refreshFrequency": "string"
  }
}

Output Schema

json

{
  "obtDesign": {
    "columns": ["object"],
    "clustering": ["string"],
    "partitioning": "object"
  },
  "buildStrategy": "object",
  "refreshConfig": "object",
  "estimatedQueryImprovement": "percentage"
}

Target Processes

•OBT Creation
•BI Dashboard Development
•Query Optimization

Usage Guidelines

•Analyze source models and relationships
•Document common query patterns
•Define platform and constraints
•Balance column count with query needs

Best Practices

•Include only columns needed for known query patterns
•Use appropriate clustering for common filter columns
•Partition by date for time-series analysis
•Schedule refreshes based on source update frequency
•Monitor query performance and adjust design