Skip to content

Search query SLA minimum replicas#

Reliability · AI Search · Rule · 2021_06 · Important

Use a minimum of 2 replicas to receive an SLA for index queries.

Description#

AI Search (Previously known as Cognitive Search) services support indexing and querying. Indexing is the process of loading content into the service to make it searchable. Querying is the process where a client searches for content by sending queries to the index.

AI Search supports a configurable number of replicas. Having multiple replicas allows queries and index updates to load balance across multiple replicas.

To receive a Service Level Agreement (SLA) for Search index queries a minimum of 2 replicas is required.

Recommendation#

Consider increasing the number of replicas to a minimum of 2 to receive an SLA on index query requests.

Examples#

Configure with Azure template#

To deploy AI Search services that pass this rule:

  • Set the properties.replicaCount property to a minimum of 2.

For example:

Azure Template snippet
{
  "type": "Microsoft.Search/searchServices",
  "apiVersion": "2022-09-01",
  "name": "[parameters('name')]",
  "location": "[parameters('location')]",
  "identity": {
    "type": "SystemAssigned"
  },
  "sku": {
    "name": "standard"
  },
  "properties": {
    "replicaCount": 3,
    "partitionCount": 1,
    "hostingMode": "default"
  }
}

Configure with Bicep#

To deploy AI Search services that pass this rule:

  • Set the properties.replicaCount property to a minimum of 2.

For example:

Azure Bicep snippet
resource search 'Microsoft.Search/searchServices@2022-09-01' = {
  name: name
  location: location
  identity: {
    type: 'SystemAssigned'
  }
  sku: {
    name: 'standard'
  }
  properties: {
    replicaCount: 3
    partitionCount: 1
    hostingMode: 'default'
  }
}

Comments