Microsoft Azure Machine Learning

문서 > 통합 > Microsoft Azure Machine Learning

개요

Azure Machine Learning 서비스는 개발자와 데이터 과학자에게 기계 학습을 구축하고, 훈련하며, 더 빨리 배포하는 데 도움이 되도록 다양한 생산적인 경험을 제공하는 서비스입니다. Datadog를 사용해 Azure Machine Learning 성능과 내 애플리케이션 및 인프라스트럭처 컨텍스트 내 활용도를 모니터링할 수 있습니다.

Azure Machine Learning 메트릭을 얻으면 다음을 할 수 있습니다.

모델 배포 및 실행 상태 횟수 추적
기계 학습 노드 활용도 모니터링
성능 대비 비용 최적화

설정

설치

아직 설정하지 않았다면, 먼저 Microsoft Azure 통합을 설정하세요. 그 외 다른 설치 단계는 없습니다.

수집한 데이터

메트릭

azure.machinelearningservices_workspaces.completed_runs (gauge)	The number of runs completed successfully for this workspace. Shown as operation
azure.machinelearningservices_workspaces.started_runs (gauge)	The number of runs started for this workspace. Shown as operation
azure.machinelearningservices_workspaces.failed_runs (gauge)	The number of runs failed for this workspace. Shown as operation
azure.machinelearningservices_workspaces.model_register_succeeded (gauge)	The number of model registrations that succeeded in this workspace.
azure.machinelearningservices_workspaces.model_register_failed (gauge)	The number of model registrations that failed in this workspace.
azure.machinelearningservices_workspaces.model_deploy_started (gauge)	The number of model deployments started in this workspace.
azure.machinelearningservices_workspaces.model_deploy_succeeded (gauge)	The number of model deployments that succeeded in this workspace.
azure.machinelearningservices_workspaces.moddel_deploy_failed (gauge)	The number of model deployments that failed in this workspace.
azure.machinelearningservices_workspaces.total_nodes (gauge)	The number of total nodes. This total includes some of Active Nodes, Idle Nodes, Unusable Nodes, Premepted Nodes, Leaving Nodes. Shown as node
azure.machinelearningservices_workspaces.active_nodes (gauge)	The number of Acitve nodes. These are the nodes which are actively running a job. Shown as node
azure.machinelearningservices_workspaces.idle_nodes (gauge)	The number of idle nodes. Idle nodes are the nodes which are not running any jobs but can accept new job if available. Shown as node
azure.machinelearningservices_workspaces.unusable_nodes (gauge)	The number of unusable nodes. Unusable nodes are not functional due to some unresolvable issue. Azure will recycle these nodes. Shown as node
azure.machinelearningservices_workspaces.preempted_nodes (gauge)	The number of preempted nodes. These nodes are the low priority nodes which are taken away from the available node pool. Shown as node
azure.machinelearningservices_workspaces.leaving_nodes (gauge)	The number of leaving nodes. Leaving nodes are the nodes which just finished processing a job and will go to Idle state. Shown as node
azure.machinelearningservices_workspaces.total_cores (gauge)	The number of total cores. Shown as core
azure.machinelearningservices_workspaces.active_cores (gauge)	The number of active cores. Shown as core
azure.machinelearningservices_workspaces.idle_cores (gauge)	The number of idle cores. Shown as core
azure.machinelearningservices_workspaces.unusable_cores (gauge)	The number of unusable cores. Shown as core
azure.machinelearningservices_workspaces.preempted_cores (gauge)	The number of preempted cores. Shown as core
azure.machinelearningservices_workspaces.leaving_cores (gauge)	The number of leaving cores. Shown as core
azure.machinelearningservices_workspaces.quota_utilization_percentage (gauge)	The percent of quota utilized. Shown as percent
azure.machinelearningservices_workspaces.cpuutilization (gauge)	CPU utilization Shown as percent
azure.machinelearningservices_workspaces.gpuutilization (gauge)	GPU utilization Shown as percent