0
# Azure Machine Learning Compute Management
1

2
A Python client library for managing Azure Machine Learning Compute resources through the Azure Resource Manager (ARM) APIs. This library enables developers to programmatically create, configure, and manage operationalization clusters for Azure Machine Learning services, including support for Kubernetes orchestration, ACS (Azure Container Service) integration, and system services management.
3

4
## Package Information
5

6
- **Package Name**: azure-mgmt-machinelearningcompute
7
- **Language**: Python
8
- **Installation**: `pip install azure-mgmt-machinelearningcompute`
9

10
## Core Imports
11

12
```python
13
from azure.mgmt.machinelearningcompute import MachineLearningComputeManagementClient
14
```
15

16
Common imports for models and types:
17

18
```python
19
from azure.mgmt.machinelearningcompute.models import (
20
    OperationalizationCluster,
21
    ClusterType,
22
    OperationStatus,
23
    StorageAccountProperties,
24
    ContainerRegistryProperties,
25
    AcsClusterProperties
26
)
27
```
28

29
## Basic Usage
30

31
```python
32
from azure.identity import DefaultAzureCredential
33
from azure.mgmt.machinelearningcompute import MachineLearningComputeManagementClient
34
from azure.mgmt.machinelearningcompute.models import (
35
    OperationalizationCluster,
36
    ClusterType,
37
    StorageAccountProperties
38
)
39

40
# Initialize the client
41
credential = DefaultAzureCredential()
42
subscription_id = "your-subscription-id"
43
client = MachineLearningComputeManagementClient(credential, subscription_id)
44

45
# Create a simple local cluster
46
cluster_params = OperationalizationCluster(
47
    location="eastus",
48
    cluster_type=ClusterType.local,
49
    description="My ML compute cluster",
50
    storage_account=StorageAccountProperties(
51
        resource_id="/subscriptions/.../storageAccounts/mystorage"
52
    )
53
)
54

55
# Create the cluster
56
operation = client.operationalization_clusters.create_or_update(
57
    resource_group_name="myresourcegroup",
58
    cluster_name="mycluster",
59
    parameters=cluster_params
60
)
61

62
# Wait for completion (long-running operation)
63
cluster = operation.result()
64
print(f"Cluster created: {cluster.name}")
65

66
# List clusters in the resource group
67
clusters = client.operationalization_clusters.list_by_resource_group("myresourcegroup")
68
for cluster in clusters:
69
    print(f"Found cluster: {cluster.name}")
70
```
71

72
## Architecture
73

74
The library follows Azure SDK for Python conventions and provides:
75

76
- **Management Client**: Primary entry point (`MachineLearningComputeManagementClient`) that coordinates operations
77
- **Operations Groups**: Specialized operation handlers for different resource types
78
  - `operationalization_clusters`: Cluster lifecycle management
79
  - `machine_learning_compute`: General compute operations
80
- **Models**: Data structures representing Azure resources and their properties
81
- **Long-running Operations**: Asynchronous operations using Azure polling patterns
82
- **Authentication**: Integration with Azure Identity for secure access
83

84
## Capabilities
85

86
### Client Management
87

88
Core client initialization, configuration, and authentication for accessing Azure Machine Learning Compute resources.
89

90
```python { .api }
91
class MachineLearningComputeManagementClient:
92
    def __init__(self, credentials, subscription_id: str, base_url: str = None): ...
93
    
94
class MachineLearningComputeManagementClientConfiguration:
95
    def __init__(self, credentials, subscription_id: str, base_url: str = None): ...
96
```
97

98
[Client Management](./client-management.md)
99

100
### Operationalization Cluster Operations
101

102
Complete lifecycle management of operationalization clusters including creation, updates, deletion, credential management, and system services.
103

104
```python { .api }
105
class OperationalizationClustersOperations:
106
    def create_or_update(self, resource_group_name: str, cluster_name: str, parameters: OperationalizationCluster) -> AzureOperationPoller: ...
107
    def get(self, resource_group_name: str, cluster_name: str) -> OperationalizationCluster: ...
108
    def delete(self, resource_group_name: str, cluster_name: str, delete_all: bool = None) -> AzureOperationPoller: ...
109
    def list_keys(self, resource_group_name: str, cluster_name: str) -> OperationalizationClusterCredentials: ...
110
```
111

112
[Cluster Operations](./cluster-operations.md)
113

114
### System Operations
115

116
System-level operations for managing available API operations and system services within clusters.
117

118
```python { .api }
119
class MachineLearningComputeOperations:
120
    def list_available_operations(self) -> AvailableOperations: ...
121
```
122

123
[System Operations](./system-operations.md)
124

125
### Data Models and Types
126

127
Comprehensive data structures representing clusters, configurations, credentials, and enumerations used throughout the API.
128

129
```python { .api }
130
class OperationalizationCluster(Resource):
131
    def __init__(self, location: str, cluster_type: ClusterType, **kwargs): ...
132

133
class StorageAccountProperties:
134
    def __init__(self, resource_id: str = None): ...
135

136
class ClusterType(Enum):
137
    acs = "ACS"
138
    local = "Local"
139
```
140

141
[Models and Types](./models-types.md)
142

143
## Types
144

145
```python { .api }
146
from typing import Union, List, Dict, Optional
147
from msrestazure.azure_operation import AzureOperationPoller
148
from msrest.pipeline import ClientRawResponse
149

150
# Core enums
151
ClusterType = Union["ACS", "Local"]
152
OperationStatus = Union["Unknown", "Updating", "Creating", "Deleting", "Succeeded", "Failed", "Canceled"]
153
OrchestratorType = Union["Kubernetes", "None"]
154
SystemServiceType = Union["None", "ScoringFrontEnd", "BatchFrontEnd"]
155
AgentVMSizeTypes = Union[str]  # Extensive list of Azure VM sizes
156
Status = Union["Enabled", "Disabled"]
157
UpdatesAvailable = Union["Yes", "No"]
158
```