Health status monitoring
Monitor the real-time health of your cloud resources to promptly address any anomalies. For more information, see Alibaba Cloud Health Status.
On the Alibaba Cloud Health Status page, you can view the real-time service status for each region and subscribe to RSS feeds for incident notifications.
CloudMonitor
PAI integrates with Alibaba Cloud CloudMonitor, a free service for real-time monitoring of your cloud resources and internet applications. CloudMonitor tracks the operational status of cloud services, resource usage of ECS instances, website availability, and business-related faults.
Enable this feature manually. For more information, see Cloud service monitoring.
One-click alerts for key metrics
CloudMonitor lets you enable alerts for multiple key PAI metrics with a single click to efficiently build a comprehensive alerting system and gain full visibility into your Alibaba Cloud resource usage and business operations. For more information, see Enable the initiative alert feature.
Custom alert rules
Create custom dashboards to collect and visualize your monitoring data. For more information, see Manage the monitoring charts of a custom dashboard.
You can set an alert rule for each metric and receive notifications through phone calls, SMS messages, emails, DingTalk chatbots, or the Alibaba Cloud app.
You can also use an alert blacklist to suppress notifications for specific metrics. For more information, see Manage blacklist policies.
Cloud Config
PAI integrates with Alibaba Cloud Cloud Config, a free-to-try service that provides centralized tracking of your cloud resource configuration history and compliance auditing. This service helps you automate compliance monitoring and ensure the continuous compliance of your infrastructure.
Resource configuration tracking
Cloud Config tracks operations from your Alibaba Cloud account and all associated RAM user accounts. It records resource configuration changes by default every 10 minutes.
MLPS 2.0 compliance pre-check
Enable the MLPS 2.0 pre-check feature with a single click to let Cloud Config continuously monitor your resource compliance. You can also download pre-check reports to submit to inspection agencies.
Audit data query and analysis
You can deliver resource configuration history and non-compliance events to a specified Logstore in Simple Log Service (SLS). You can then use SLS to query and analyze all your log data from a central location. For more information, see Deliver resource data to a Logstore in Simple Log Service.
ActionTrail
PAI integrates with Alibaba Cloud ActionTrail to provide unified management of your cloud resource operation logs. ActionTrail records user logins and resource access events across your cloud account to enable security analysis, intrusion detection, resource change tracking, and compliance auditing.
ActionTrail can record log data for actions taken through the Alibaba Cloud console, APIs, and developer tools. For a list of supported cloud services and audit events, see Audit events of ECS.
By default, ActionTrail tracks and retains events for the last 90 days. To store logs for a longer period, you must create a trail to deliver events to Simple Log Service (SLS) or Object Storage Service (OSS). For more information, see Getting Started.
After you deliver events to SLS or OSS, you can query and analyze them from the SLS or OSS console. For more information, see Query events in the Simple Log Service or OSS console.
If you need to track historical events, submit a ticket to request access.
Workspace event center
PAI provides a workspace-level event center where you can create event rules. These rules let you track and monitor the status of DLC and workflow tasks, or to trigger downstream events based on changes in the admission status of model versions in AI Asset Management. Supported event sources include workflow tasks, DLC tasks, and model events. Notifications can be sent via DingTalk, voice calls, or email. For more information, see Workspace notification.
TensorBoard task analysis
You can launch TensorBoard in Machine Learning Designer to visualize model training analysis reports.
-
Machine Learning Designer: Use TensorBoard to visualize analytical reports.
-
DLC: TensorBoard.