你当前正在访问 Microsoft Azure Global Edition 技术文档网站。 如果需要访问由世纪互联运营的 Microsoft Azure 中国技术文档网站,请访问 https://docs.azure.cn

Microsoft.MachineLearningServices 工作区 2023-02-01-preview

Bicep 资源定义

可以使用目标操作部署工作区资源类型:

有关每个 API 版本中已更改属性的列表,请参阅 更改日志

资源格式

若要创建 Microsoft.MachineLearningServices/workspaces 资源,请将以下 Bicep 添加到模板。

resource symbolicname 'Microsoft.MachineLearningServices/workspaces@2023-02-01-preview' = {
  identity: {
    type: 'string'
    userAssignedIdentities: {
      {customized property}: {}
    }
  }
  kind: 'string'
  location: 'string'
  name: 'string'
  properties: {
    allowPublicAccessWhenBehindVnet: bool
    applicationInsights: 'string'
    containerRegistry: 'string'
    description: 'string'
    discoveryUrl: 'string'
    encryption: {
      identity: {
        userAssignedIdentity: 'string'
      }
      keyVaultProperties: {
        identityClientId: 'string'
        keyIdentifier: 'string'
        keyVaultArmId: 'string'
      }
      status: 'string'
    }
    featureStoreSettings: {
      computeRuntime: {
        sparkRuntimeVersion: 'string'
      }
      offlineStoreConnectionName: 'string'
      onlineStoreConnectionName: 'string'
    }
    friendlyName: 'string'
    hbiWorkspace: bool
    imageBuildCompute: 'string'
    keyVault: 'string'
    primaryUserAssignedIdentity: 'string'
    publicNetworkAccess: 'string'
    serviceManagedResourcesSettings: {
      cosmosDb: {
        collectionsThroughput: int
      }
    }
    sharedPrivateLinkResources: [
      {
        name: 'string'
        properties: {
          groupId: 'string'
          privateLinkResourceId: 'string'
          requestMessage: 'string'
          status: 'string'
        }
      }
    ]
    softDeleteRetentionInDays: int
    storageAccount: 'string'
    systemDatastoresAuthMode: 'string'
    v1LegacyMode: bool
  }
  sku: {
    capacity: int
    family: 'string'
    name: 'string'
    size: 'string'
    tier: 'string'
  }
  tags: {
    {customized property}: 'string'
  }
}

属性值

微软.机器学习服务/工作区

名字 描述 价值
标识 资源的标识。 ManagedServiceIdentity
字符串
位置 指定资源的位置。 字符串
名字 资源名称 字符串

约束:
模式 = ^[a-zA-Z0-9][a-zA-Z0-9_-]{2,32}$ (必需)
性能 机器学习工作区的属性。 WorkspaceProperties (工作区属性)
SKU 工作区的 SKU。 SKU
标签 资源标记 标记名称和值的字典。 请参阅模板 中的 标记

ComputeRuntimeD设置为

名字 描述 价值
sparkRuntimeVersion 字符串

CosmosDb设置

名字 描述 价值
collections吞吐量 cosmosdb 数据库中集合的吞吐量 整数 (int)

EncryptionKeyVaultProperties

名字 描述 价值
identityClientId 将来使用 - 将用于访问密钥保管库的标识的客户端 ID。 字符串
keyIdentifier 键标识符 用于访问加密密钥的密钥保管库 URI。 string (必需)
keyVaultArmId 客户拥有加密密钥的 keyVault 的 ArmId。 string (必需)

EncryptionProperty 属性

名字 描述 价值
标识 将用于访问密钥保管库的标识进行静态加密。 IdentityForCmk
keyVaultProperties 客户密钥保管库属性。 EncryptionKeyVaultProperties (必需)
状态 指示是否为工作区启用加密。 “Disabled”
“Enabled”(必需)

FeatureStoreSettings (功能商店设置)

名字 描述 价值
computeRuntime 的 ComputeRuntimeD设置为
offlineStoreConnectionName (离线存储连接名称) 字符串
onlineStoreConnectionName (在线商店连接名称) 字符串

IdentityForCmk

名字 描述 价值
userAssignedIdentity 将用于访问客户管理的密钥保管库的用户分配标识的 ArmId 字符串

ManagedServiceIdentity

名字 描述 价值
类型 托管服务标识的类型(其中允许 SystemAssigned 和 UserAssigned 类型)。 “None”
“SystemAssigned”
“SystemAssigned,UserAssigned”
“UserAssigned”(必需)
用户分配的标识 与资源关联的用户分配标识集。 userAssignedIdentities 字典密钥将为 ARM 资源 ID,格式为“/subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.ManagedIdentity/userAssignedIdentities/{identityName}”。 字典值可以是请求中的空对象({})。 UserAssignedIdentities

ServiceManagedResourcesSettings

名字 描述 价值
cosmosDb 数据库 服务托管 cosmosdb 帐户的设置。 CosmosDb设置

SharedPrivateLinkResource

名字 描述 价值
名字 专用链接的唯一名称。 字符串
性能 资源属性。 SharedPrivateLinkResourceProperty

SharedPrivateLinkResourceProperty

名字 描述 价值
组 ID 专用链接资源组 ID。 字符串
privateLinkResourceId 专用链接链接到的资源 ID。 字符串
请求消息 请求消息。 字符串
状态 指示连接是否已被服务的所有者批准/拒绝/删除。 “已批准”
“Disconnected”
“Pending”
“已拒绝”
“Timeout”

SKU

名字 描述 价值
产能 如果 SKU 支持横向扩展/缩减,则应包含容量整数。 如果资源无法横向扩展/缩小,则可能会省略此范围。 整数 (int)
家庭 如果服务具有不同代的硬件(对于同一 SKU,则可以在此处捕获)。 字符串
名字 SKU 的名称。 例 - P3。 它通常是字母+数字代码 string (必需)
大小 SKU 大小。 当名称字段是层和其他一些值的组合时,这将是独立的代码。 字符串
分层 如果服务具有多个层,但在 PUT 上不需要,则需要资源提供程序实现此字段。 “Basic”
“Free”
“Premium”
“Standard”

UserAssignedIdentities

名字 描述 价值

UserAssignedIdentity

名字 描述 价值

WorkspaceProperties (工作区属性)

名字 描述 价值
allowPublicAccessWhenBehindVnet 指示在 VNet 后面是否允许公共访问的标志。 布尔
applicationInsights 与此工作区关联的 application Insights 的 ARM ID。 字符串
containerRegistry 与此工作区关联的容器注册表的 ARM ID。 字符串
描述 此工作区的说明。 字符串
发现网址 用于标识机器学习试验服务的区域终结点的发现服务的 URL 字符串
加密 Azure ML 工作区的加密设置。 EncryptionProperty 属性
featureStoreSettings 功能存储类型工作区的设置。 FeatureStoreSettings (功能商店设置)
友好名称 此工作区的友好名称。 此名称在可变中 字符串
hbi工作区 用于向工作区中的 HBI 数据发出信号并减少服务收集的诊断数据的标志 布尔
imageBuildCompute 映像生成的计算名称 字符串
密钥保险库 与此工作区关联的密钥保管库的 ARM ID。 创建工作区后,无法更改此更改 字符串
primaryUserAssignedIdentity 表示工作区标识的用户分配的标识资源 ID。 字符串
公用网络访问 是否允许来自公用网络的请求。 “Disabled”
“Enabled”
serviceManagedResourcesSettings 服务托管资源设置。 ServiceManagedResourcesSettings
sharedPrivateLink资源 此工作区中共享的专用链接资源列表。 SharedPrivateLinkResource[]
softDeleteRetentionInDays (英语) 工作区软删除后的保留时间(以天为单位)。 整数 (int)
存储帐户 与此工作区关联的存储帐户的 ARM ID。 创建工作区后,无法更改此更改 字符串
systemDatastoresAuthMode 用于访问工作区的系统数据存储的身份验证模式 字符串
v1LegacyMode 启用v1_legacy_mode可能会阻止你使用 v2 API 提供的功能。 布尔

工作区标签

名字 描述 价值

用法示例

Azure 验证模块

以下 Azure 验证模块 可用于部署此资源类型。

模块 描述
机器学习服务工作区 机器学习服务工作区的 AVM 资源模块

Azure 快速入门示例

以下 Azure 快速入门模板 包含用于部署此资源类型的 Bicep 示例。

Bicep 文件 描述
Azure AI Foundry 基本设置 这组模板演示了如何使用基本设置设置 Azure AI Foundry,这意味着启用公共 Internet 访问、用于加密的 Microsoft 托管密钥以及 AI 资源的 Microsoft 托管身份配置。
Azure AI Foundry 基本设置 这组模板演示了如何使用基本设置设置 Azure AI Foundry,这意味着启用公共 Internet 访问、用于加密的 Microsoft 托管密钥以及 AI 资源的 Microsoft 托管身份配置。
Azure AI Foundry 网络受限 这组模板演示了如何在禁用专用链接和出口的情况下设置 Azure AI Foundry,使用 Microsoft 管理的密钥进行加密,并使用 Microsoft 管理的标识配置为 AI 资源。
具有 Microsoft Entra ID 身份验证的 Azure AI Foundry 这组模板演示了如何为依赖资源(如 Azure AI 服务和 Azure 存储)设置 Azure AI Foundry 和 Microsoft Entra ID 身份验证。
Azure AI Studio 基本设置 此模板集演示如何使用基本设置设置设置 Azure AI Studio,这意味着启用了公共 Internet 访问、Microsoft托管密钥进行加密和 AI 资源的Microsoft托管标识配置。
Azure AI Studio 网络受限 此模板集演示如何使用禁用专用链接和出口设置 Azure AI Studio,并使用Microsoft托管密钥进行加密和 AI 资源的Microsoft托管标识配置。
Azure 机器学习端到端安全设置 此 Bicep 模板集演示如何在安全设置中端到端设置 Azure 机器学习。 此参考实现包括工作区、计算群集、计算实例和附加的专用 AKS 群集。
Azure 机器学习端到端安全设置(旧版) 此 Bicep 模板集演示如何在安全设置中端到端设置 Azure 机器学习。 此参考实现包括工作区、计算群集、计算实例和附加的专用 AKS 群集。
基本代理设置 API 密钥 此模板集演示如何使用 API 密钥身份验证为 AI 服务/AOAI 连接设置基本设置 Azure AI 代理服务。 代理使用完全由Microsoft管理的多租户搜索和存储资源。 无法查看或控制这些基础 Azure 资源。
基本代理设置标识 此模板集演示如何使用 AI 服务/AOAI 连接的托管标识身份验证设置基本设置 Azure AI 代理服务。 代理使用完全由Microsoft管理的多租户搜索和存储资源。 无法查看或控制这些基础 Azure 资源。
创建具有专用 IP 地址的 AKS 计算目标 此模板使用专用 IP 地址在给定的 Azure 机器学习服务工作区中创建 AKS 计算目标。
创建 Azure 机器学习服务工作区 此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 此配置描述了开始使用 Azure 机器学习所需的最少资源集。
创建 Azure 机器学习服务工作区(CMK) 此部署模板指定如何使用加密密钥创建具有服务端加密的 Azure 机器学习工作区。
创建 Azure 机器学习服务工作区(CMK) 此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 该示例演示如何使用客户管理的加密密钥配置 Azure 机器学习进行加密。
创建 Azure 机器学习服务工作区(旧版) 此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 此配置描述了在网络隔离设置中开始使用 Azure 机器学习所需的资源集。
创建 Azure 机器学习服务工作区(vnet) 此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 此配置描述了在网络隔离设置中开始使用 Azure 机器学习所需的资源集。
使用托管虚拟网络部署 Secure AI Foundry 此模板创建具有强大网络和标识安全限制的安全 Azure AI Foundry 环境。
使用用户托管标识 网络安全代理 此模板集演示如何使用用户托管标识身份验证为 AI 服务/AOAI 连接设置虚拟网络隔离的 Azure AI 代理服务,以及专用网络链接,以将代理连接到安全数据。
标准代理设置 此模板集演示如何使用标准设置设置设置 Azure AI 代理服务,这意味着已启用项目/中心连接的托管标识身份验证和公共 Internet 访问。 代理使用客户拥有的单租户搜索和存储资源。 通过此设置,可以完全控制这些资源并查看这些资源,但会根据使用情况产生成本。

ARM 模板资源定义

可以使用目标操作部署工作区资源类型:

有关每个 API 版本中已更改属性的列表,请参阅 更改日志

资源格式

若要创建 Microsoft.MachineLearningServices/workspaces 资源,请将以下 JSON 添加到模板。

{
  "type": "Microsoft.MachineLearningServices/workspaces",
  "apiVersion": "2023-02-01-preview",
  "name": "string",
  "identity": {
    "type": "string",
    "userAssignedIdentities": {
      "{customized property}": {
      }
    }
  },
  "kind": "string",
  "location": "string",
  "properties": {
    "allowPublicAccessWhenBehindVnet": "bool",
    "applicationInsights": "string",
    "containerRegistry": "string",
    "description": "string",
    "discoveryUrl": "string",
    "encryption": {
      "identity": {
        "userAssignedIdentity": "string"
      },
      "keyVaultProperties": {
        "identityClientId": "string",
        "keyIdentifier": "string",
        "keyVaultArmId": "string"
      },
      "status": "string"
    },
    "featureStoreSettings": {
      "computeRuntime": {
        "sparkRuntimeVersion": "string"
      },
      "offlineStoreConnectionName": "string",
      "onlineStoreConnectionName": "string"
    },
    "friendlyName": "string",
    "hbiWorkspace": "bool",
    "imageBuildCompute": "string",
    "keyVault": "string",
    "primaryUserAssignedIdentity": "string",
    "publicNetworkAccess": "string",
    "serviceManagedResourcesSettings": {
      "cosmosDb": {
        "collectionsThroughput": "int"
      }
    },
    "sharedPrivateLinkResources": [
      {
        "name": "string",
        "properties": {
          "groupId": "string",
          "privateLinkResourceId": "string",
          "requestMessage": "string",
          "status": "string"
        }
      }
    ],
    "softDeleteRetentionInDays": "int",
    "storageAccount": "string",
    "systemDatastoresAuthMode": "string",
    "v1LegacyMode": "bool"
  },
  "sku": {
    "capacity": "int",
    "family": "string",
    "name": "string",
    "size": "string",
    "tier": "string"
  },
  "tags": {
    "{customized property}": "string"
  }
}

属性值

微软.机器学习服务/工作区

名字 描述 价值
apiVersion API 版本 “2023-02-01-预览”
标识 资源的标识。 ManagedServiceIdentity
字符串
位置 指定资源的位置。 字符串
名字 资源名称 字符串

约束:
模式 = ^[a-zA-Z0-9][a-zA-Z0-9_-]{2,32}$ (必需)
性能 机器学习工作区的属性。 WorkspaceProperties (工作区属性)
SKU 工作区的 SKU。 SKU
标签 资源标记 标记名称和值的字典。 请参阅模板 中的 标记
类型 资源类型 “Microsoft.MachineLearningServices/workspaces”

ComputeRuntimeD设置为

名字 描述 价值
sparkRuntimeVersion 字符串

CosmosDb设置

名字 描述 价值
collections吞吐量 cosmosdb 数据库中集合的吞吐量 整数 (int)

EncryptionKeyVaultProperties

名字 描述 价值
identityClientId 将来使用 - 将用于访问密钥保管库的标识的客户端 ID。 字符串
keyIdentifier 键标识符 用于访问加密密钥的密钥保管库 URI。 string (必需)
keyVaultArmId 客户拥有加密密钥的 keyVault 的 ArmId。 string (必需)

EncryptionProperty 属性

名字 描述 价值
标识 将用于访问密钥保管库的标识进行静态加密。 IdentityForCmk
keyVaultProperties 客户密钥保管库属性。 EncryptionKeyVaultProperties (必需)
状态 指示是否为工作区启用加密。 “Disabled”
“Enabled”(必需)

FeatureStoreSettings (功能商店设置)

名字 描述 价值
computeRuntime 的 ComputeRuntimeD设置为
offlineStoreConnectionName (离线存储连接名称) 字符串
onlineStoreConnectionName (在线商店连接名称) 字符串

IdentityForCmk

名字 描述 价值
userAssignedIdentity 将用于访问客户管理的密钥保管库的用户分配标识的 ArmId 字符串

ManagedServiceIdentity

名字 描述 价值
类型 托管服务标识的类型(其中允许 SystemAssigned 和 UserAssigned 类型)。 “None”
“SystemAssigned”
“SystemAssigned,UserAssigned”
“UserAssigned”(必需)
用户分配的标识 与资源关联的用户分配标识集。 userAssignedIdentities 字典密钥将为 ARM 资源 ID,格式为“/subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.ManagedIdentity/userAssignedIdentities/{identityName}”。 字典值可以是请求中的空对象({})。 UserAssignedIdentities

ServiceManagedResourcesSettings

名字 描述 价值
cosmosDb 数据库 服务托管 cosmosdb 帐户的设置。 CosmosDb设置

SharedPrivateLinkResource

名字 描述 价值
名字 专用链接的唯一名称。 字符串
性能 资源属性。 SharedPrivateLinkResourceProperty

SharedPrivateLinkResourceProperty

名字 描述 价值
组 ID 专用链接资源组 ID。 字符串
privateLinkResourceId 专用链接链接到的资源 ID。 字符串
请求消息 请求消息。 字符串
状态 指示连接是否已被服务的所有者批准/拒绝/删除。 “已批准”
“Disconnected”
“Pending”
“已拒绝”
“Timeout”

SKU

名字 描述 价值
产能 如果 SKU 支持横向扩展/缩减,则应包含容量整数。 如果资源无法横向扩展/缩小,则可能会省略此范围。 整数 (int)
家庭 如果服务具有不同代的硬件(对于同一 SKU,则可以在此处捕获)。 字符串
名字 SKU 的名称。 例 - P3。 它通常是字母+数字代码 string (必需)
大小 SKU 大小。 当名称字段是层和其他一些值的组合时,这将是独立的代码。 字符串
分层 如果服务具有多个层,但在 PUT 上不需要,则需要资源提供程序实现此字段。 “Basic”
“Free”
“Premium”
“Standard”

UserAssignedIdentities

名字 描述 价值

UserAssignedIdentity

名字 描述 价值

WorkspaceProperties (工作区属性)

名字 描述 价值
allowPublicAccessWhenBehindVnet 指示在 VNet 后面是否允许公共访问的标志。 布尔
applicationInsights 与此工作区关联的 application Insights 的 ARM ID。 字符串
containerRegistry 与此工作区关联的容器注册表的 ARM ID。 字符串
描述 此工作区的说明。 字符串
发现网址 用于标识机器学习试验服务的区域终结点的发现服务的 URL 字符串
加密 Azure ML 工作区的加密设置。 EncryptionProperty 属性
featureStoreSettings 功能存储类型工作区的设置。 FeatureStoreSettings (功能商店设置)
友好名称 此工作区的友好名称。 此名称在可变中 字符串
hbi工作区 用于向工作区中的 HBI 数据发出信号并减少服务收集的诊断数据的标志 布尔
imageBuildCompute 映像生成的计算名称 字符串
密钥保险库 与此工作区关联的密钥保管库的 ARM ID。 创建工作区后,无法更改此更改 字符串
primaryUserAssignedIdentity 表示工作区标识的用户分配的标识资源 ID。 字符串
公用网络访问 是否允许来自公用网络的请求。 “Disabled”
“Enabled”
serviceManagedResourcesSettings 服务托管资源设置。 ServiceManagedResourcesSettings
sharedPrivateLink资源 此工作区中共享的专用链接资源列表。 SharedPrivateLinkResource[]
softDeleteRetentionInDays (英语) 工作区软删除后的保留时间(以天为单位)。 整数 (int)
存储帐户 与此工作区关联的存储帐户的 ARM ID。 创建工作区后,无法更改此更改 字符串
systemDatastoresAuthMode 用于访问工作区的系统数据存储的身份验证模式 字符串
v1LegacyMode 启用v1_legacy_mode可能会阻止你使用 v2 API 提供的功能。 布尔

工作区标签

名字 描述 价值

用法示例

Azure 快速入门模板

以下 Azure 快速入门模板 部署此资源类型。

模板 描述
Azure AI Foundry 基本设置

部署到 Azure
这组模板演示了如何使用基本设置设置 Azure AI Foundry,这意味着启用公共 Internet 访问、用于加密的 Microsoft 托管密钥以及 AI 资源的 Microsoft 托管身份配置。
Azure AI Foundry 基本设置

部署到 Azure
这组模板演示了如何使用基本设置设置 Azure AI Foundry,这意味着启用公共 Internet 访问、用于加密的 Microsoft 托管密钥以及 AI 资源的 Microsoft 托管身份配置。
Azure AI Foundry 网络受限

部署到 Azure
这组模板演示了如何在禁用专用链接和出口的情况下设置 Azure AI Foundry,使用 Microsoft 管理的密钥进行加密,并使用 Microsoft 管理的标识配置为 AI 资源。
具有 Microsoft Entra ID 身份验证的 Azure AI Foundry

部署到 Azure
这组模板演示了如何为依赖资源(如 Azure AI 服务和 Azure 存储)设置 Azure AI Foundry 和 Microsoft Entra ID 身份验证。
Azure AI Studio 基本设置

部署到 Azure
此模板集演示如何使用基本设置设置设置 Azure AI Studio,这意味着启用了公共 Internet 访问、Microsoft托管密钥进行加密和 AI 资源的Microsoft托管标识配置。
Azure AI Studio 网络受限

部署到 Azure
此模板集演示如何使用禁用专用链接和出口设置 Azure AI Studio,并使用Microsoft托管密钥进行加密和 AI 资源的Microsoft托管标识配置。
Azure 机器学习端到端安全设置

部署到 Azure
此 Bicep 模板集演示如何在安全设置中端到端设置 Azure 机器学习。 此参考实现包括工作区、计算群集、计算实例和附加的专用 AKS 群集。
Azure 机器学习端到端安全设置(旧版)

部署到 Azure
此 Bicep 模板集演示如何在安全设置中端到端设置 Azure 机器学习。 此参考实现包括工作区、计算群集、计算实例和附加的专用 AKS 群集。
Azure 机器学习工作区

部署到 Azure
此模板创建新的 Azure 机器学习工作区,以及加密的存储帐户、KeyVault 和 Applications Insights 日志记录
基本代理设置 API 密钥

部署到 Azure
此模板集演示如何使用 API 密钥身份验证为 AI 服务/AOAI 连接设置基本设置 Azure AI 代理服务。 代理使用完全由Microsoft管理的多租户搜索和存储资源。 无法查看或控制这些基础 Azure 资源。
基本代理设置标识

部署到 Azure
此模板集演示如何使用 AI 服务/AOAI 连接的托管标识身份验证设置基本设置 Azure AI 代理服务。 代理使用完全由Microsoft管理的多租户搜索和存储资源。 无法查看或控制这些基础 Azure 资源。
创建包含多个数据集的 AML 工作区 & 数据存储

部署到 Azure
此模板创建包含多个数据集 & 数据存储的 Azure 机器学习工作区。
创建具有专用 IP 地址的 AKS 计算目标

部署到 Azure
此模板使用专用 IP 地址在给定的 Azure 机器学习服务工作区中创建 AKS 计算目标。
创建 Azure 机器学习服务工作区

部署到 Azure
此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 此配置描述了开始使用 Azure 机器学习所需的最少资源集。
创建 Azure 机器学习服务工作区(CMK)

部署到 Azure
此部署模板指定如何使用加密密钥创建具有服务端加密的 Azure 机器学习工作区。
创建 Azure 机器学习服务工作区(CMK)

部署到 Azure
此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 该示例演示如何使用客户管理的加密密钥配置 Azure 机器学习进行加密。
创建 Azure 机器学习服务工作区(旧版)

部署到 Azure
此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 此配置描述了在网络隔离设置中开始使用 Azure 机器学习所需的资源集。
创建 Azure 机器学习服务工作区(vnet)

部署到 Azure
此部署模板指定 Azure 机器学习工作区及其关联的资源,包括 Azure Key Vault、Azure 存储、Azure Application Insights 和 Azure 容器注册表。 此配置描述了在网络隔离设置中开始使用 Azure 机器学习所需的资源集。
使用托管虚拟网络部署 Secure AI Foundry

部署到 Azure
此模板创建具有强大网络和标识安全限制的安全 Azure AI Foundry 环境。
使用用户托管标识 网络安全代理

部署到 Azure
此模板集演示如何使用用户托管标识身份验证为 AI 服务/AOAI 连接设置虚拟网络隔离的 Azure AI 代理服务,以及专用网络链接,以将代理连接到安全数据。
标准代理设置

部署到 Azure
此模板集演示如何使用标准设置设置设置 Azure AI 代理服务,这意味着已启用项目/中心连接的托管标识身份验证和公共 Internet 访问。 代理使用客户拥有的单租户搜索和存储资源。 通过此设置,可以完全控制这些资源并查看这些资源,但会根据使用情况产生成本。

Terraform (AzAPI 提供程序)资源定义

可以使用目标操作部署工作区资源类型:

  • 资源组

有关每个 API 版本中已更改属性的列表,请参阅 更改日志

资源格式

若要创建 Microsoft.MachineLearningServices/workspaces 资源,请将以下 Terraform 添加到模板。

resource "azapi_resource" "symbolicname" {
  type = "Microsoft.MachineLearningServices/workspaces@2023-02-01-preview"
  name = "string"
  parent_id = "string"
  identity {
    type = "string"
    identity_ids = [
      "string"
    ]
  }
  location = "string"
  tags = {
    {customized property} = "string"
  }
  body = {
    kind = "string"
    properties = {
      allowPublicAccessWhenBehindVnet = bool
      applicationInsights = "string"
      containerRegistry = "string"
      description = "string"
      discoveryUrl = "string"
      encryption = {
        identity = {
          userAssignedIdentity = "string"
        }
        keyVaultProperties = {
          identityClientId = "string"
          keyIdentifier = "string"
          keyVaultArmId = "string"
        }
        status = "string"
      }
      featureStoreSettings = {
        computeRuntime = {
          sparkRuntimeVersion = "string"
        }
        offlineStoreConnectionName = "string"
        onlineStoreConnectionName = "string"
      }
      friendlyName = "string"
      hbiWorkspace = bool
      imageBuildCompute = "string"
      keyVault = "string"
      primaryUserAssignedIdentity = "string"
      publicNetworkAccess = "string"
      serviceManagedResourcesSettings = {
        cosmosDb = {
          collectionsThroughput = int
        }
      }
      sharedPrivateLinkResources = [
        {
          name = "string"
          properties = {
            groupId = "string"
            privateLinkResourceId = "string"
            requestMessage = "string"
            status = "string"
          }
        }
      ]
      softDeleteRetentionInDays = int
      storageAccount = "string"
      systemDatastoresAuthMode = "string"
      v1LegacyMode = bool
    }
    sku = {
      capacity = int
      family = "string"
      name = "string"
      size = "string"
      tier = "string"
    }
  }
}

属性值

微软.机器学习服务/工作区

名字 描述 价值
标识 资源的标识。 ManagedServiceIdentity
字符串
位置 指定资源的位置。 字符串
名字 资源名称 字符串

约束:
模式 = ^[a-zA-Z0-9][a-zA-Z0-9_-]{2,32}$ (必需)
性能 机器学习工作区的属性。 WorkspaceProperties (工作区属性)
SKU 工作区的 SKU。 SKU
标签 资源标记 标记名称和值的字典。
类型 资源类型 “Microsoft.MachineLearningServices/workspaces@2023-02-01-preview”

ComputeRuntimeD设置为

名字 描述 价值
sparkRuntimeVersion 字符串

CosmosDb设置

名字 描述 价值
collections吞吐量 cosmosdb 数据库中集合的吞吐量 整数 (int)

EncryptionKeyVaultProperties

名字 描述 价值
identityClientId 将来使用 - 将用于访问密钥保管库的标识的客户端 ID。 字符串
keyIdentifier 键标识符 用于访问加密密钥的密钥保管库 URI。 string (必需)
keyVaultArmId 客户拥有加密密钥的 keyVault 的 ArmId。 string (必需)

EncryptionProperty 属性

名字 描述 价值
标识 将用于访问密钥保管库的标识进行静态加密。 IdentityForCmk
keyVaultProperties 客户密钥保管库属性。 EncryptionKeyVaultProperties (必需)
状态 指示是否为工作区启用加密。 “Disabled”
“Enabled”(必需)

FeatureStoreSettings (功能商店设置)

名字 描述 价值
computeRuntime 的 ComputeRuntimeD设置为
offlineStoreConnectionName (离线存储连接名称) 字符串
onlineStoreConnectionName (在线商店连接名称) 字符串

IdentityForCmk

名字 描述 价值
userAssignedIdentity 将用于访问客户管理的密钥保管库的用户分配标识的 ArmId 字符串

ManagedServiceIdentity

名字 描述 价值
类型 托管服务标识的类型(其中允许 SystemAssigned 和 UserAssigned 类型)。 “None”
“SystemAssigned”
“SystemAssigned,UserAssigned”
“UserAssigned”(必需)
用户分配的标识 与资源关联的用户分配标识集。 userAssignedIdentities 字典密钥将为 ARM 资源 ID,格式为“/subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.ManagedIdentity/userAssignedIdentities/{identityName}”。 字典值可以是请求中的空对象({})。 UserAssignedIdentities

ServiceManagedResourcesSettings

名字 描述 价值
cosmosDb 数据库 服务托管 cosmosdb 帐户的设置。 CosmosDb设置

SharedPrivateLinkResource

名字 描述 价值
名字 专用链接的唯一名称。 字符串
性能 资源属性。 SharedPrivateLinkResourceProperty

SharedPrivateLinkResourceProperty

名字 描述 价值
组 ID 专用链接资源组 ID。 字符串
privateLinkResourceId 专用链接链接到的资源 ID。 字符串
请求消息 请求消息。 字符串
状态 指示连接是否已被服务的所有者批准/拒绝/删除。 “已批准”
“Disconnected”
“Pending”
“已拒绝”
“Timeout”

SKU

名字 描述 价值
产能 如果 SKU 支持横向扩展/缩减,则应包含容量整数。 如果资源无法横向扩展/缩小,则可能会省略此范围。 整数 (int)
家庭 如果服务具有不同代的硬件(对于同一 SKU,则可以在此处捕获)。 字符串
名字 SKU 的名称。 例 - P3。 它通常是字母+数字代码 string (必需)
大小 SKU 大小。 当名称字段是层和其他一些值的组合时,这将是独立的代码。 字符串
分层 如果服务具有多个层,但在 PUT 上不需要,则需要资源提供程序实现此字段。 “Basic”
“Free”
“Premium”
“Standard”

UserAssignedIdentities

名字 描述 价值

UserAssignedIdentity

名字 描述 价值

WorkspaceProperties (工作区属性)

名字 描述 价值
allowPublicAccessWhenBehindVnet 指示在 VNet 后面是否允许公共访问的标志。 布尔
applicationInsights 与此工作区关联的 application Insights 的 ARM ID。 字符串
containerRegistry 与此工作区关联的容器注册表的 ARM ID。 字符串
描述 此工作区的说明。 字符串
发现网址 用于标识机器学习试验服务的区域终结点的发现服务的 URL 字符串
加密 Azure ML 工作区的加密设置。 EncryptionProperty 属性
featureStoreSettings 功能存储类型工作区的设置。 FeatureStoreSettings (功能商店设置)
友好名称 此工作区的友好名称。 此名称在可变中 字符串
hbi工作区 用于向工作区中的 HBI 数据发出信号并减少服务收集的诊断数据的标志 布尔
imageBuildCompute 映像生成的计算名称 字符串
密钥保险库 与此工作区关联的密钥保管库的 ARM ID。 创建工作区后,无法更改此更改 字符串
primaryUserAssignedIdentity 表示工作区标识的用户分配的标识资源 ID。 字符串
公用网络访问 是否允许来自公用网络的请求。 “Disabled”
“Enabled”
serviceManagedResourcesSettings 服务托管资源设置。 ServiceManagedResourcesSettings
sharedPrivateLink资源 此工作区中共享的专用链接资源列表。 SharedPrivateLinkResource[]
softDeleteRetentionInDays (英语) 工作区软删除后的保留时间(以天为单位)。 整数 (int)
存储帐户 与此工作区关联的存储帐户的 ARM ID。 创建工作区后,无法更改此更改 字符串
systemDatastoresAuthMode 用于访问工作区的系统数据存储的身份验证模式 字符串
v1LegacyMode 启用v1_legacy_mode可能会阻止你使用 v2 API 提供的功能。 布尔

工作区标签

名字 描述 价值

用法示例

Azure 验证模块

以下 Azure 验证模块 可用于部署此资源类型。

模块 描述
机器学习服务工作区 机器学习服务工作区的 AVM 资源模块