Stack Monitoring provides discovery and monitoring of applications and infrastructure. These features are licensed in two editions: Standard and Enterprise Editions. These editions can be enabled on a per resource basis or across an entire compartment. Features associated with each edition are described below.

Stack Monitoring also provides monitoring and alarm management for GPU infrastructure and its running workloads. These features are licensed using the Enterprise Edition for GPU Infrastructure. Details are also described below.

Stack Monitoring Standard Edition provides discovery and monitoring of applications and its underlying stack deployed on-premises and cloud (OCI or 3rd party clouds). These discovered entities are referred to as ‘resources.’ Monitored resources include applications such as E-Business Suite and PeopleSoft, and application tech stack components such as WebLogic Servers, Tomcat servers, Oracle databases, SQL Server databases and hosts. Features include:

• Discovery – discovery of an application and its topology
• Monitoring and troubleshooting UI
  • Enterprise Health and Alarm Summary
    • Interactive UI to monitor status and health across an entire fleet
    • Review open alarms and drilldown to further triage.
    • Identify performance outliers and dynamically correlate any 2 metrics in charts for streamlined troubleshooting.
    • Drilldown to resource homepages for further triage.
    • Create customizable enterprise views based on personas.
  • Resource home pages
    • Monitor status, performance and alarms for a specific resource.
    • Customize performance charts.
    • Use topology to review health across the stack and navigate to related resources.
    • Use application Stack Views to monitor health and stack performance in a single view.

Stack Monitoring Enterprise Edition provides advanced monitoring and alarm management in addition to all Standard Edition features. Features include:

• All features in Standard Edition
• Advanced Monitoring
  • Anomaly detection – auto-detection of performance anomalies based on applying machine learning (ML) algorithms to performance data
  • Metric Extensions - creation, test and deployment of custom metrics based on user-defined scripts
  • Process monitoring –status and performance monitoring of any hosted application based on its running processes
• Alarm Management
  • Monitoring Templates – resource-oriented management of alarm definitions in a single UI
  • Maintenance Windows – resource-oriented management of alarm (notification) suppressions in support of maintenance periods
• Extensibility
  • Extends monitoring scope by monitoring other services and infrastructure types.
  • OCI Services – monitor any OCI Service by importing it as a resource into Stack Monitoring. Once imported, include it in any application topology for enhanced end-to-end monitoring visibility across the app stack.
  • Prometheus integration – monitor any app or infrastructure by ingesting its metrics exposed in Prometheus format. Monitoring UI (homepage) will automatically be created for this new resource.
  • collectd and Telegraf integration – monitor a wide variety of apps and infrastructure by integrating with collectd and Telegraf. Monitoring UI (homepages) will automatically be created for these new resources.

Stack Monitoring Enterprise Edition for GPU Infrastructure provides monitoring and alarm management for GPU infrastructure and its running workloads. This license is based on the GPU Type and number of GPUs. All Stack Monitoring features are included with this license. Monitoring workloads running on GPU infrastructure are also included with this license and not charged separately. Features include:

• All Advanced Monitoring and Extensibility features in Enterprise Edition
• Discovery – discovery of GPU hosts and network topology
• Monitoring and troubleshooting UI
  • Enterprise Health and Alarms for Host GPU
    • Interactive UI to monitor status and health across a GPU host fleet or GPU Cluster Network
    • Review open alarms and drilldown to further triage.
    • Identify performance outliers, e.g. GPUs with highest average latency or GPUs with highest ECC errors
    • Dynamically correlate any two metrics in charts (e.g. GPU Power and GPU Temperature) for streamlined troubleshooting
    • Drilldown to resource homepages for further triage.
    • Create customizable enterprise views based on personas.
  • Resource home pages
    • Monitor status, performance and alarms for all hosts and GPUs that are part of a Cluster Network.
    • Use topology to review health across the stack and navigate throughout the cluster network, network block, local block, hosts and GPUs.
    • Use ML-based anomaly detection to identify performance outliers.
  • Workload monitoring using Process Sets
• Alarm Management
  • Monitoring Templates – manage all alarm settings across all hosts and GPUs in a single UI.
  • Maintenance Windows – suppress alarm notifications for hosts and GPUs.