Skip to main content

Create an Azure AI Search Service

Azure AI Search is an information retrieval platform for the enterprise. It supports traditional search and conversational, AI-driven search for “chat with your data” experiences. The easiest way to create a search service is through the Azure portal, covered in this guide.

Before You Start

Some properties are fixed for the lifetime of the search service. Decide on these before creating your service:
PropertyDescription
NameBecomes part of the URL endpoint (must be unique)
RegionDetermines data residency and feature availability
TierDetermines infrastructure, service limits, and billing
Compute typeStandard VMs (recommended) or confidential VMs

Prerequisites

  • An Azure subscription (start free trial)
  • Permissions to create resources in your subscription
Each Azure subscription can have one free Azure AI Search service, which is perfect for evaluation and learning.

Create Your Search Service

1

Sign in to Azure Portal

Navigate to the Azure portal and sign in with your Azure account
2

Create a Resource

Click Create a resource in the upper-left corner, then search for Azure AI Search
3

Choose Subscription

Select the Azure subscription where you want to create the search service
4

Set Resource Group

Choose an existing resource group or create a new one:
  • Resource groups help organize related resources
  • They enable cost tracking and management
  • Make cleanup easier when testing
5

Name Your Service

Enter a unique service name:
  • Must be 2-60 characters
  • Use only lowercase letters, digits, and dashes
  • Cannot start or end with dashes
  • Cannot contain consecutive dashes
  • Becomes part of your endpoint: https://your-service-name.search.windows.net
6

Select Region

Choose the region where your service will be hosted. Consider:
  • Proximity to your users and data sources
  • Feature availability (some features require specific regions)
  • Capacity availability (some regions may be at capacity)
  • Agentic retrieval region requirements
7

Choose Pricing Tier

Select your pricing tier based on your needs. See Pricing Tiers below for details.
8

Select Compute Type

  • Standard (recommended): Standard Azure VMs with data encrypted at rest and in transit
  • Confidential: Hardware-based trusted execution environment (10% surcharge)
9

Review and Create

Review your configuration and click Create to deploy the service

Pricing Tiers

Azure AI Search offers multiple pricing tiers:
Free Tier
  • One per subscription
  • Perfect for learning and evaluation
  • 50 MB storage limit
  • 3 indexes maximum
  • No SLA
  • May be deleted if inactive for extended periods
Complete most quickstarts and tutorials on the Free tier.
Services created after April 3, 2024 have larger partitions and higher vector quotas at every billable tier.

Naming Guidelines

Valid Service Names

mysearchservice
company-search-prod
search2024

Invalid Service Names

MySearchService    // uppercase not allowed
-search-           // cannot start/end with dash
search--service    // consecutive dashes not allowed
company.search     // dots not allowed
Include the region in your service name (e.g., myservice-westus) to make multi-region management easier.

Region Selection

Checklist for Choosing a Region

Check if your target region supports:
  • Your desired pricing tier
  • AI enrichment capabilities
  • Integrated vectorization
  • Multimodal search
  • Agentic retrieval (preview)
See region availability by tier
  • Keep services in the same region to minimize latency
  • No bandwidth charges for same-region services
  • Consider compliance and data sovereignty requirements
For AI enrichment with Azure AI services:
  • Key-based connections require both services in the same region
  • Keyless connections (preview) allow different regions
  • Check Azure Vision availability for multimodal embeddings
For high availability:
  • Create services in multiple regions
  • Use 2+ replicas per service for availability zones
  • Example: East US and West US for North America coverage

Configure Authentication

By default, new services use API key authentication. For better security, enable role-based access control (RBAC).
1

Navigate to Keys

In your search service, go to Settings > Keys
2

Choose Authentication Method

Select one of:
  • API Key: Traditional key-based authentication
  • Role-based access control: Azure AD authentication (recommended)
  • Both: Allow both methods during migration
3

Assign Roles

For RBAC, assign these roles to users or applications:
  • Search Service Contributor: Manage service and indexes
  • Search Index Data Contributor: Add/update/delete documents
  • Search Index Data Reader: Query indexes (read-only)
API keys provide full access to your service. Rotate keys regularly and consider switching to RBAC for production.

Scale Your Service

After deployment, you can scale your service to meet demand.

Replicas and Partitions

  • Replicas: Copies of your index for high availability and query throughput
    • Minimum 2 for read-only SLA
    • Minimum 3 for read/write SLA
    • Distribute across availability zones automatically
  • Partitions: Divide your index for storage and parallel processing
    • More storage capacity
    • Higher indexing throughput
    • Parallel query execution

Scaling Example

1

Go to Scale Settings

Navigate to Settings > Scale in your search service
2

Adjust Replicas

Use the slider to increase or decrease replicas
3

Adjust Partitions

Use the slider to increase or decrease partitions
4

Review Cost

Check the estimated monthly cost before applying changes
Scaling is only available on billable tiers. Free tier services cannot be scaled.

Compute Type Selection

Standard Compute (Default)

  • Standard Azure VMs
  • Data encrypted at rest and in transit
  • Suitable for most workloads
  • Base pricing
Use when:
  • Building typical search applications
  • Cost optimization is important
  • Standard security is sufficient

Confidential Compute

  • Azure confidential computing VMs
  • Hardware-based trusted execution environment
  • Data protected in use (during processing)
  • 10% surcharge
  • Limited regional availability
  • Some features restricted
Use when:
  • Processing highly sensitive data
  • Compliance requires data-in-use protection
  • Regulatory requirements mandate confidential computing
Confidential compute has limited regional availability and disables certain features. Review data in use documentation before selecting.

Monitor Your Service

Once deployed, monitor service health and usage:
  1. Service Dashboard: View overview metrics
  2. Metrics: Query latency, throttling, index size
  3. Logs: Diagnostic logs for troubleshooting
  4. Alerts: Set up notifications for issues

Cost Management

Optimize your search service costs:

Right-Size Resources

Start small and scale up based on actual usage metrics

Monitor Usage

Use Azure Cost Management to track spending

Optimize Indexing

Schedule indexer runs during off-peak hours

Review Replicas

Scale down replicas when high availability isn’t needed

When to Add a Second Service

Consider creating additional services for:
  • Multi-region deployment: Reduce latency for global users
  • Disaster recovery: Ensure service availability during regional outages
  • Environment isolation: Separate dev, test, and production
  • Multi-tenant architectures: Isolate customer data
  • Workload separation: Different SLAs for different applications
You cannot separate indexing and querying workloads. Both operations run on the same service.

Request Additional Quota

Azure subscriptions have limits on the number of search services:
  1. Navigate to Quotas service in Azure portal
  2. Filter by Search services
  3. Select your subscription and region
  4. Click Request adjustment
  5. Enter new limit (must exceed current)
  6. Submit request
Most requests are approved within 24 hours. Large quota increases (30+ services) may take up to one month.

Alternative Creation Methods

Besides the Azure portal, you can create services using:

Azure PowerShell

New-AzSearchService -ResourceGroupName "myResourceGroup" `
    -Name "mysearchservice" `
    -Location "West US" `
    -Sku "Standard"

Azure CLI

az search service create \
    --name mysearchservice \
    --resource-group myResourceGroup \
    --location westus \
    --sku Standard

ARM Template

{
  "type": "Microsoft.Search/searchServices",
  "apiVersion": "2023-11-01",
  "name": "mysearchservice",
  "location": "westus",
  "sku": {
    "name": "standard"
  },
  "properties": {
    "replicaCount": 1,
    "partitionCount": 1
  }
}

Bicep

resource searchService 'Microsoft.Search/searchServices@2023-11-01' = {
  name: 'mysearchservice'
  location: 'westus'
  sku: {
    name: 'standard'
  }
  properties: {
    replicaCount: 1
    partitionCount: 1
  }
}

Next Steps

Create an Index

Build your first search index

Import Data

Load data from Azure services

Security

Configure authentication and access control

Monitoring

Set up monitoring and alerts

Build docs developers (and LLMs) love