Navigate the entity graph
The entity graph provides a visual representation of the relationships between infrastructure and application components in your IT environment.
The knowledge graph generates an entity graph by analyzing metrics and discovering application and infrastructure components and the relationships between them. The knowledge graph updates the graph every minute. The metrics that form the basis of the graph can come from various sources such as Prometheus exporters, Istio, Grafana Beyla, metrics derived from OpenTelemetry Traces, and LogQL queries on raw logs.
The entity graph shows different types of relationships. For example:
- When a node hosts a Pod, the relationship is a
HOSTS
relationship - When a service calls another service, the relationship is a
CALLS
relationship
Because the graph is continuously updated, you can visualize changes to entities and relationships. For instance, if a Pod moves from Node-1 to Node-2, depending on the selected time window, you see Node-1 or Node-2 or both Node-1 and Node-2 as a connected entity of the Pod. Similarly, if a Pod is removed, it appears greyed out.
Each entity has a quick link so that you can view all connected entities and one or more curated KPI dashboards. Additionally, the advanced search feature enables you to search for entities by name or other attributes and see their connections. You can also create a custom search expression to focus on specific parts of the architecture, allowing for a more thorough exploration and understanding of your system.
An entity might have colored rings around it. Blue (info) indicates software configuration changes on that entity. Yellow indicates warning level severity problems and red indicates critical level severity problems. When the inner circle of an entity changes color, it signifies that entities contained within the parent entity triggered an insight.
Before you begin
Before you use the entity graph, ensure you have completed the following:
- Check that your infrastructure and applications generate metrics, logs, and traces
- Activate the knowledge graph
Open the entity graph
Complete the following steps to open the entity graph.
Sign into Grafana and click Observability > Entity graph.
From the Frequently used menu, select Show all Services or Show all Nodes.
Use predefined searches
The first time you use the entity graph, by default, a large number of entities might appear. You can make your view more manageable by using the basic search feature to reduce the graph to include only those entities of interest. Using search expressions is the best way to explore the entity graph.
The entity graph includes a number of human-readable search expressions that address everyday use cases. For example, you can search for Show Service api-server
, Show all services
, Show Assertions in Namespace
, and so on.
To use basic search, begin the search expression with the word Show
and continue typing keywords or select a search value from the drop-down list.
Define advanced search
The Advanced search panel provides enhanced search functions such as searching by property conditions and displaying connected entities. These searches allow you to quickly access a group of entities with contextual information.
Because the knowledge graph continuously updates the graph with time-series data, the search results reflect the system’s state within the specified time window. Changing the time settings might result in different search results. Additionally, the search results are always filtered based on the selections made in the env and site drop-down menus.
The Advanced search feature is particularly useful for creating customized views specific to your area of responsibility.
You can save these views and easily access them for future reference.
To use Advanced search, complete the following steps:
To the right of the search field, click Advanced search.
On the Advanced search tab, build the search expression.
Navigate the entity graph
Clicking on an entity enables you to navigate to its connected entities. This navigation is possible in both the graph view and the list view. If the graph becomes too crowded, you can filter the displayed entities by clicking on the entity type in the legend.
To navigate the entity graph, complete the following steps:
Click an entity in the graph.
A menu appears showing you all connected entity types and the number of each type.
The following example shows navigating to all services associated with the Cart node.
Select a connected entity type.
The entity graph:
- Updates to show all connected entities
- Displays all connected entities in the Graph view
- Provides a Back arrow so that you can navigate back to the original entity.
You can also use the Graph view to navigate to connected entities.
To filter the entity graph, click an entity and click Only.
View and interpret insights
Each entity in the system is represented by two color-coded rings that reflect their insight status:
- The outer ring indicates the status of the entity itself.
- The inner ring indicates the status of a lower-level entity connected to the entity.
The inner ring is particularly useful because, in the case of temporary entities like Pods, the parent entities such as nodes or services are typically more significant for analysis. Therefore, observing a colored inner ring on the node or service provides valuable information.
The ring color represents the severity of the insight:
- Red indicates
CRITICAL
- Yellow indicates
WARNING
- Blue indicates for
INFO
In general, saturation insights are CRITICAL
or WARNING
, failures are usually CRITICAL
, anomalies are usually WARNING
, and amends are usually INFO
.
The following example shows the Shipping
entity, including its properties and insights. The Connected Entities section displays a colored dot that indicates whether any critical, warning, or info insights have been triggered on the connected entities linked to Shipping.
To view and interpret insights, complete the following steps:
Hover over an entity to reveal its details and the insights firing.
To add the entity to RCA workbench, click the plus sign (+).
RCA workbench provides tools for you to perform troubleshooting and root cause analysis.
You can add multiple entities to RCA workbench.
To instantly navigate to RCA workbench to perform root cause analysis, click Troubleshoot in RCA workbench.
Navigate to dashboards
Rather than building your own dashboards that you must maintain over time, the knowledge graph includes a number of dashboards that you can navigate to from the entity graph. Because the knowledge graph constantly monitors your environment, the dashboards are always available and up to date, even when there are underlying changes to your infrastructure or applications. Each entity type has its own dashboard. For example, there are dashboards for clusters, node groups, nodes, Pods, services, and so on.
If you have your own Grafana dashboards, you can integrate them into the knowledge graph.
To navigate to dashboards:
Hover over an entity and click Show KPI.
A drawer opens to the KPI dashboard.
To navigate to other dashboards, click the dashboard name at the top of the drawer.
Use a list view
In the list view, you have the ability to select properties or monitoring statuses for each entity. This provides a comprehensive overview of the elements that are automatically monitored by the knowledge graph. Resource monitoring is applicable to both services and nodes, while traffic and latency monitoring typically apply only to services.
To generate a list view, complete the following steps:
Click List.
Use the Columns checkboxes to add properties to the list view.