dataverse-schema-design

This skill provides guidance on designing Dataverse table schemas and data models. Use when users ask about "Dataverse table design", "Dataverse schema", "Dataverse relationships", "Dataverse columns", "data modeling Dataverse", "Dataverse best practices", or need help designing their data structure.

Table Design Fundamentals

Naming Conventions

• Table prefix: Use publisher prefix (e.g., new_, cr123_)
• Table names: PascalCase, singular (e.g., new_Project, new_Task)
• Column names: prefix_columnname (e.g., new_projectname, new_startdate)

Create Table with SDK

# Basic table
client.create_table("new_Project", {
    "new_ProjectName": "string",      # Text column
    "new_Description": "string",       # Multi-line text
    "new_Budget": "decimal",           # Currency/decimal
    "new_StartDate": "datetime",       # Date and time
    "new_IsActive": "bool",            # Yes/No
    "new_Priority": Priority           # Choice/enum
})

Column Types

Text Columns

# Single line (max 4000 chars)
"new_Name": "string"

# Multi-line text (memo)
"new_Description": "string"  # Will be nvarchar(max)

Number Columns

# Integer
"new_Quantity": "int"

# Decimal (with precision)
"new_Amount": "decimal"  # Default precision

# Currency (use decimal with formatting)
"new_Budget": "decimal"

Date/Time Columns

# Date and time
"new_StartDate": "datetime"

# Date only (format in app)
"new_DueDate": "datetime"

Boolean Columns

# Yes/No
"new_IsActive": "bool"
"new_IsApproved": "bool"

Choice (Picklist) Columns

from enum import IntEnum

class Status(IntEnum):
    DRAFT = 1
    SUBMITTED = 2
    APPROVED = 3
    REJECTED = 4

class Priority(IntEnum):
    LOW = 1
    MEDIUM = 2
    HIGH = 3

# Create table with choices
client.create_table("new_Request", {
    "new_Name": "string",
    "new_Status": Status,
    "new_Priority": Priority
})

Relationship Patterns

One-to-Many (1:N)

Parent table has many child records.

Account (1) ←→ (N) Contact
└── An account has many contacts

Project (1) ←→ (N) Task
└── A project has many tasks

Implementation:

• Create lookup column on child table
• Reference parent using @odata.bind

# Create contact linked to account
contact = {
    "firstname": "John",
    "lastname": "Doe",
    "parentcustomerid_account@odata.bind": f"/accounts({account_id})"
}
client.create("contact", contact)

Many-to-Many (N:N)

Records in both tables can relate to multiple records in the other.

Account (N) ←→ (N) Contact
└── Contacts can be related to multiple accounts

Note: N:N relationships require creating an intersection entity via the UI or advanced API calls.

Self-Referential

Table references itself (e.g., employee hierarchy).

Employee
├── new_ManagerId → Employee
└── Parent employee record

Schema Design Best Practices

Do's

1. Plan before creating - Design schema on paper first
2. Use meaningful names - Clear, descriptive column names
3. Add descriptions - Document purpose of each column
4. Set required fields - Enforce data quality
5. Use appropriate types - Don't store numbers as text
6. Create indexes - On frequently filtered columns
7. Use lookups - Instead of duplicating data

Don'ts

1. Don't over-normalize - Balance between normalization and performance
2. Don't use reserved names - Avoid system column names
3. Don't create wide tables - Split into related tables if >100 columns
4. Don't store calculated data - Use calculated columns instead
5. Don't ignore security - Plan field-level security early

Common Schema Patterns

Master-Detail

Account (Master)
├── new_AccountNumber
├── new_Name
└── Contacts (Detail)
    ├── new_FirstName
    ├── new_LastName
    └── _parentcustomerid_value (FK)

Status Workflow

new_Request
├── new_Name
├── new_Status (Draft → Submitted → Approved/Rejected)
├── new_SubmittedOn
├── new_ApprovedBy
└── new_ApprovedOn

Audit Trail

new_Order
├── new_OrderNumber
├── new_Status
├── createdon (system)
├── createdby (system)
├── modifiedon (system)
└── modifiedby (system)

Adding Columns to Existing Tables

# Add new columns
client.create_columns("new_Project", {
    "new_CompletionPercentage": "int",
    "new_ActualEndDate": "datetime"
})

# Remove columns
client.delete_columns("new_Project", ["new_OldColumn"])

References

• See references/table-design.md for detailed patterns
• See references/relationships.md for relationship examples
• See references/performance.md for optimization tips