Azure AI Search

Azure AI Search (formerly Azure Cognitive Search) enables full-text search capabilities across your Business Central data. This guide covers configuring Cloud Replicator to use Azure AI Search as a destination.

Overview

Azure AI Search indexes your Business Central data for powerful search capabilities:

AI Search Concept	Business Central Mapping
Index	Business Central table name (sanitized, lowercase)
Document	Business Central record
Document Key	SystemId (GUID, required)
Fields	Business Central fields with appropriate types

When to Use Azure AI Search

Azure AI Search is ideal for:

Use Case	Benefit
Product search	Full-text search across item descriptions
Customer lookup	Find customers by partial name, address, etc.
Document search	Search posted documents by content
Knowledge bases	Create searchable help or FAQ systems

Prerequisites

Before configuring Azure AI Search:

Azure Subscription: Active Azure subscription
Search Service: Azure AI Search service created
Admin Key: Admin API key (not query key)

Creating an Azure AI Search Service

Step 1: Create Service in Azure Portal

Sign in to the Azure Portal (https://portal.azure.com)
Click "Create a resource"
Search for "Azure AI Search" and select it
Click "Create"

Step 2: Configure Basic Settings

Setting	Recommendation
Subscription	Your subscription
Resource group	Create new or use existing
Service name	Unique name (becomes part of URL)
Location	Same region as Business Central
Pricing tier	Start with Basic or Standard

Pricing Tier Selection

Tier	Documents	Index Size	Use Case
Free	10,000	50 MB	Testing only
Basic	1 million	2 GB	Small datasets
Standard	15 million per partition	25 GB per partition	Production
Standard 2/3	Higher limits	Higher limits	Large scale

Step 3: Complete Creation

Click "Review + create"
Review settings and click "Create"
Wait for deployment to complete

Step 4: Get Admin Key

Navigate to your Search service
Go to "Keys" under Settings
Copy one of the Admin keys (not Query keys)

Configuring Cloud Replicator

Setup Steps

Search for "Cloud Replicator Setup" in Business Central
Locate the Azure AI Search section
Enter the following:

Field	Value
Azure AI Search Service Name	Your service name (just the name, not URL)
Azure AI Search Key	An Admin API key

Testing the Connection

After entering credentials:

Create a table mapping (e.g., Customer or Item)
Select "Azure AI Search" as the destination
Ensure "SystemId as RowKey" is enabled (required for AI Search)
Run a manual replication
Check the Replication Log for success
Verify index in Azure Portal > Search service > Indexes

Important Limitations

Azure AI Search has specific requirements and limitations:

SystemId Required

Azure AI Search requires SystemId as the document key:

"SystemId as RowKey" is automatically enabled for AI Search destinations
Cannot use primary key fields as the document key

Per Day Not Supported

The "Per Day" (series tables) feature is not available for Azure AI Search:

This option is automatically disabled for AI Search destinations
Use other destinations for time-series data

Deletions Not Supported

Record deletion tracking is not currently supported for Azure AI Search:

Deleted records remain in the search index
Manually remove deleted documents or rebuild indexes periodically

Company Name Field Not Added

Unlike the Azure Table and Cosmos DB destinations, the Field Name for Company configured in Cloud Replicator Setup is not written to Azure AI Search documents. As a result, you cannot filter the search index by company name. If you need to distinguish records by company, include a company-specific field via Field Actions, or use a separate index per company.

Index Structure

Index Creation

Cloud Replicator automatically:

Creates an index for each replicated table
Defines field types based on Business Central field types
Configures searchable, filterable, and sortable attributes

Field Type Mapping

Business Central Type	AI Search Type	Searchable
Text, Code	Edm.String	Yes
Date	Edm.DateTimeOffset	No
Integer, Duration	Edm.Int32	No
BigInteger	Edm.Int64	No
Decimal	Edm.Double	No
Boolean	Edm.Boolean	No
DateTime, Time, GUID, and any other type	Edm.String	Yes
Blob (as text)	Edm.String	Yes
Blob (as Base64)	Edm.String	Yes

The document key field is SystemId (Edm.String, marked as the index key).

Index Naming

Index names are derived from table names:

Spaces and special characters are removed
Converted to lowercase
Must be unique within the service

Example: "Sales Header" becomes "salesheader"

Querying the Search Index

Azure Portal

Navigate to your Search service
Select "Search explorer"
Select your index
Enter search queries

Example queries:

// Simple search
Adatum

// Field-specific search
Name:Adatum

// Filter by a field value
$filter=City eq 'Atlanta'

// Combined search and filter
searchText=Corporation&$filter=City eq 'Atlanta'

REST API

POST https://[service].search.windows.net/indexes/customers/docs/search?api-version=2024-07-01
Content-Type: application/json
api-key: [query-key]

{
  "search": "Adatum",
  "filter": "City eq 'Atlanta'",
  "select": "Name, Address, City",
  "top": 10
}

SDK Access

// C# Example
var client = new SearchClient(
    new Uri("https://[service].search.windows.net"),
    "customers",
    new AzureKeyCredential("[query-key]"));

var results = await client.SearchAsync<Customer>("Adatum",
    new SearchOptions
    {
        Filter = "City eq 'Atlanta'",
        Select = { "Name", "Address", "City" }
    });

Search Features

Full-Text Search

All text fields are searchable by default:

Searches across all searchable fields
Supports wildcards, fuzzy matching, and phrases
Automatic language analysis

Filtering

Non-searchable fields support filtering:

Exact match filters
Range queries for dates and numbers
Logical operators (and, or, not)

Enable facets for category-based navigation:

Group results by field values
Show counts per category
Build interactive search interfaces

Sorting

Sort results by any sortable field:

Ascending or descending order
Multiple sort criteria

Performance Considerations

Index Size

Monitor index size in Azure Portal
Upgrade tier if approaching limits
Exclude unnecessary fields to reduce size

Query Performance

Use filters to narrow results
Limit selected fields with $select
Use $top to limit result count

Replication Performance

AI Search indexing is moderately fast
Large initial loads may take time
Subsequent replications are incremental

Cost Management

Pricing Components

Component	Description
Service tier	Monthly charge based on tier
Document count	Counted against tier limits
Storage	Included in tier, counted against limits
Queries	Included (no per-query charge)

Cost Optimization

Choose appropriate tier for your data volume
Replicate only tables needed for search
Exclude unnecessary fields
Consider Basic tier for small datasets

Troubleshooting

Connection Errors

Verify service name is correct (without .search.windows.net)
Ensure you're using an Admin key, not Query key
Check service firewall settings

Index Creation Fails

Verify Admin key has create permissions
Check for invalid characters in table names
Review error in Replication Log

Documents Not Appearing

Wait for indexing to complete (can take minutes)
Check Replication Log for errors
Verify table mapping is configured correctly
Confirm "SystemId as RowKey" is enabled

Search Returns No Results

Verify documents were indexed
Check search syntax
Ensure field names match (use sanitized names)
Review filter syntax

Security Best Practices

Key Management

Use Admin keys only for indexing/management
Use Query keys for search applications
Rotate keys periodically
Store keys securely

Network Security

Configure IP firewall rules
Use private endpoints for production
Restrict Admin key access

Data Security

Index only necessary data
Exclude sensitive fields using Field Actions
Review what data is searchable