Azure Cloud Deployment

Chalk’s feature platform with best-in-class developer experience enables machine learning teams to focus on building the unique products and models that make their business stand out. Chalk provides a feature store so that you can deploy production machine learning pipelines for real time data in minutes.

Chalk is both a framework and a platform — developers can write code using familiar Python packages, and deploy their feature and data pipeline definitions to Chalk’s platform. In the Customer Cloud deployment, Chalk runs & administers its platform on the customer’s cloud account. Chalk’s managed infrastructure then executes the customer defined pipelines to compute feature data for machine learning applications. Chalk then serves this data back to customer applications for online inference and to customer data teams for training set generation.

​

Azure Service Principal Setup

Chalk requires certain permissions to manage infrastructure in your Azure account. Typically, Chalk recommends:

• Create a Service Principal for Chalk in your Azure Active Directory
• Assign the custom role with the permissions listed below
• Configure AKS RBAC for Kubernetes resource management
• Set up Key Vault access for secrets management

​

Required Azure Permissions

Chalk recommends creating a custom role in your subscription with the following permissions granted. At a high level, Chalk needs the ability to provision the key components of your infrastructure:

​

Core Infrastructure Permissions

json{
"properties": {
"roleName": "Chalk Platform Manager",
"description": "Custom role for Chalk platform deployment and management",
"permissions": [
    {
    "actions": [
    "Microsoft.Resources/subscriptions/resourceGroups/read",
    "Microsoft.Resources/subscriptions/resourceGroups/write",
    "Microsoft.Resources/subscriptions/resourceGroups/delete",
    "Microsoft.Resources/deployments/*",
    "Microsoft.Resources/tags/*",

              "Microsoft.Compute/virtualMachines/*",
              "Microsoft.Compute/virtualMachineScaleSets/*",
              "Microsoft.Compute/disks/*",
              "Microsoft.Compute/snapshots/*",
              "Microsoft.Compute/images/*",
              "Microsoft.Compute/galleries/*",
              "Microsoft.Compute/proximityPlacementGroups/*",

              "Microsoft.Network/virtualNetworks/*",
              "Microsoft.Network/publicIPAddresses/*",
              "Microsoft.Network/loadBalancers/*",
              "Microsoft.Network/applicationGateways/*",
              "Microsoft.Network/networkSecurityGroups/*",
              "Microsoft.Network/routeTables/*",
              "Microsoft.Network/privateDnsZones/*",
              "Microsoft.Network/privateEndpoints/*",
              "Microsoft.Network/networkInterfaces/*",

              "Microsoft.ContainerService/managedClusters/*",
              "Microsoft.ContainerService/locations/*",
              "Microsoft.ContainerService/operations/read",

              "Microsoft.ContainerRegistry/registries/*",
              "Microsoft.ContainerRegistry/locations/*",

              "Microsoft.Storage/storageAccounts/*",
              "Microsoft.Storage/locations/*",

              "Microsoft.KeyVault/vaults/*",
              "Microsoft.KeyVault/deletedVaults/*",
              "Microsoft.KeyVault/locations/*",
              "Microsoft.KeyVault/operations/read",

              "Microsoft.Insights/components/*",
              "Microsoft.Insights/workbooks/*",
              "Microsoft.Insights/actionGroups/*",
              "Microsoft.Insights/alertRules/*",
              "Microsoft.Insights/metricAlerts/*",
              "Microsoft.Insights/scheduledQueryRules/*",
              "Microsoft.Insights/diagnosticSettings/*",

              "Microsoft.OperationalInsights/workspaces/*",
              "Microsoft.OperationalInsights/solutions/*",
              "Microsoft.OperationalInsights/savedSearches/*",

              "Microsoft.Cache/redis/*",
              "Microsoft.Cache/locations/*",

              "Microsoft.EventHub/namespaces/*",
              "Microsoft.EventHub/locations/*",

              "Microsoft.Sql/servers/*",
              "Microsoft.Sql/managedInstances/*",
              "Microsoft.Sql/locations/*",

              "Microsoft.DocumentDB/databaseAccounts/*",
              "Microsoft.DocumentDB/locations/*",

              "Microsoft.Authorization/roleAssignments/read",
              "Microsoft.Authorization/roleAssignments/write",
              "Microsoft.Authorization/roleAssignments/delete",
              "Microsoft.Authorization/roleDefinitions/read",

              "Microsoft.ManagedIdentity/userAssignedIdentities/*"
            ],
            "notActions": [],
            "dataActions": [
              "Microsoft.KeyVault/vaults/secrets/*",
              "Microsoft.KeyVault/vaults/keys/*",
              "Microsoft.KeyVault/vaults/certificates/read",

              "Microsoft.Storage/storageAccounts/blobServices/containers/*",
              "Microsoft.Storage/storageAccounts/fileServices/fileshares/*",
              "Microsoft.Storage/storageAccounts/queueServices/queues/*",
              "Microsoft.Storage/storageAccounts/tableServices/tables/*"
            ],
            "notDataActions": []
          }
        ],
        "assignableScopes": [
          "/subscriptions/{subscription-id}"
        ]
    }
}

Azure allows you to assign these permissions at various levels of granularity:

• Subscription level (recommended): /subscriptions/{subscription-id}
• Resource Group level: /subscriptions/{subscription-id}/resourceGroups/{resource-group-name}
• Individual Resource level: /subscriptions/{subscription-id}/resourceGroups/{resource-group-name}/providers/{resource-provider}/{resource-name}

We suggest following the normal practice of full subscription separation from other resources for simplicity and security. However, if your organization requires more granular access control, Chalk can work within specific Resource Groups by updating the assignableScopes accordingly.

​

What are these permissions for?

• Azure Kubernetes Service (AKS): Container orchestration for running feature engineering workloads
• Azure Container Registry (ACR): Container image management for deployments
• Azure Storage: Object storage for feature data and artifacts
• Azure SQL Database: Relational database for online and offline storage
• Azure Cosmos DB: NoSQL database for online storage
• Azure Cache for Redis: In-memory caching for feature serving
• Azure Event Hubs: Event streaming for real-time data ingestion
• Azure Service Bus: Message queuing for asynchronous processing
• Azure Key Vault: Secrets management for data source credentials
• Azure Monitor: Logging and metrics collection
• Azure Application Insights: Application performance monitoring

​

Kubernetes RBAC Configuration

Chalk requires the following Kubernetes roles to manage the resources in your AKS cluster:

​

Cluster-Level Role

kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: chalk-cluster-management-role
rules:
- apiGroups:
  - ""
  resources:
  - nodes # required to track usage
  verbs:
  - get
  - list
  - watch
# For allowing the web UI to manage cluster scaling
- apiGroups:
  - "cluster-autoscaler.kubernetes.io"
  resources:
  - nodepools
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - delete
- apiGroups:
  - ""
  resources:
  - persistentvolumes
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - watch
  - delete
- apiGroups:
  - "storage.k8s.io"
  resources:
  - storageclasses
  verbs:
  - get
  - list

​

Namespace-Level Role

kind: Role
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: chalk-management-role
  namespace: <namespace> # replace <namespace> with your actual namespace
rules:
- apiGroups:
  - ""
  resources:
  - configmaps
  - pods
  - services
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - watch
# For support w/ debugging & rendering logs in the dashboard.
- apiGroups:
  - ""
  resources:
  - pods/log
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - "extensions"
  - "networking.k8s.io"
  resources:
  - ingresses
  - ingresses/status
  - ingressclasses
  verbs:
  - get
  - list
  - watch
  - delete
  - update
  - create
- apiGroups:
  - ""
  resources:
  - secrets
  verbs:
  - get
  - list
  - watch
  - create
  - patch
  - update
  - delete
- apiGroups:
  - ""
  resources:
  - events
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - "apps"
  resources:
  - replicasets
  - statefulsets
  - deployments
  - daemonsets
  verbs:
  - get
  - list
  - update
  - patch
  - create
  - delete
- apiGroups:
  - "batch"
  resources:
  - cronjobs
  - jobs
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - watch
  - delete
# For durable storage management.
- apiGroups:
  - ""
  resources:
  - persistentvolumeclaims
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - watch
  - delete
# For managing KEDA objects for autoscaling.
- apiGroups:
  - keda.sh
  resources:
  - scaledobjects
  - scaledjobs
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - watch
  - delete
# For managing PodDisruptionBudgets.
- apiGroups:
  - policy
  resources:
  - poddisruptionbudgets
  verbs:
  - get
  - list
  - create
  - update
  - patch
  - watch
# OPTIONAL: For showing thread dumps & profiling for batch backfills + some support use-cases.
- apiGroups:
  - ""
  resources:
  - pods/exec
  verbs:
  - create
  - get
# OPTIONAL: For support with debugging k8s rbac, grant read-only access to this namespaces' rbac configuration.
- apiGroups: ["", "rbac.authorization.k8s.io"]
  resources: ["roles", "serviceaccounts", "rolebindings"]
  verbs: ["get", "list"]

​

Workload Service Account Role

Additionally, Chalk workload service accounts require the following role to be able to manage batch workloads:

apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
name: chalk-job-reader
namespace: <namespace> # replace <namespace> with your actual namespace
rules:
- apiGroups: ["batch"]
resources: ["jobs"]
verbs: ["get", "list", "watch"]

​

Argo Workflows Role (Optional)

If you use in-cluster docker image building powered by kaniko and Argo, Chalk requires this role on the namespace where the docker image building is happening:

apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
name: argo-workflows-role
namespace: <namespace> # replace <namespace> with your actual namespace
rules:
- apiGroups: ["argoproj.io"]
  resources: ["workflows"]
  verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
- apiGroups: [""]
  resources: ["pods"]
  verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]

This role must be granted to the management service account that Chalk uses to manage the cluster, and to the workload service account that runs the docker image building. Azure Monitor and Log Analytics In order to display logs in the Chalk web UI, Chalk requires permissions to be able to read Azure Monitor logs. If you use a separate logging service, you can set up Azure Monitor Agent or Fluent Bit to send logs to Log Analytics workspace. Then, to view these logs we need the following permissions:

{
    "permissions": [
        {
            "actions": [
                "Microsoft.OperationalInsights/workspaces/read",
                "Microsoft.OperationalInsights/workspaces/query/read",
                "Microsoft.OperationalInsights/workspaces/search/read",
                "Microsoft.OperationalInsights/workspaces/sharedKeys/action",
                "Microsoft.Insights/logs/read"
            ],
            "notActions": [],
            "dataActions": [
                "Microsoft.OperationalInsights/workspaces/query/*/read",
                "Microsoft.OperationalInsights/workspaces/search/*/read"
            ],
            "notDataActions": []
        }
    ]
}