Map service dependencies and track health across your infrastructure

Prerequisites

• User role with hosts.view permission for viewing groups, dependencies, topology, and health history
• User role with hosts.manage permission for creating/modifying/deleting dependencies
• User role with host_groups.view / host_groups.manage for group CRUD and membership
• Hosts must have agents reporting network flows for auto-discovery (connection sampling must be enabled in agent settings)

Understanding service maps

The service map is built on top of host groups. Any host group can optionally become a health-tracked "service" by configuring a health calculation method. Dependencies between groups (and standalone hosts) represent network connections, either discovered automatically from flow data or defined manually.

Key concepts

Creating service nodes

To create a health-tracked group (equivalent to the old "Business Service" concept):

1. Navigate to Host Groups > Groups tab.
2. Click Create Group.
3. Fill in name, description, and select the organization.
4. Set Health Calculation to one of the available methods:

   Method	Behavior
   Weighted Average	Weighted average of member host health scores based on assigned weights.
   Worst Case	Health score equals the lowest individual member score. One unhealthy host degrades the entire group.
   Majority	Health reflects the majority status of members. Group is healthy if more than half the members are healthy.

5. Set degraded and critical thresholds (percentage scores below which the group status changes).
6. Add member hosts with weight and role.
7. Click Save.

Defining connections and dependencies

Dependencies represent network relationships between host groups and/or standalone hosts.

Manual dependency creation

1. Navigate to Host Groups > Dependencies tab.
2. Click Add Dependency.
3. Select the source group or host and the target group or host.
4. Optionally specify the port and protocol for the connection.
5. Click Save.

Manual dependencies have maximum confidence (1.0) and are never overwritten or decayed by auto-discovery.

Auto-discovered dependencies

The dependency discovery engine automatically infers connections from network flow data:

1. Agents collect TCP/UDP connections every 300 seconds.
2. Flow data is submitted to the backend and stored for analysis.
3. The discovery engine resolves destination IPs to known hosts.
4. It infers host-level and group-level dependencies from the observed traffic patterns.

Dependency lifecycle

Impact analysis

The topology graph enables upstream and downstream impact analysis by following dependency edges.

Using the topology graph

1. Navigate to Host Groups > Dependencies tab.
2. The graph view shows all host groups as nodes and dependencies as edges.
3. Click a node to see its details: health score, status, member count, and connected dependencies.
4. Click an edge to see connection details: port, protocol, confidence, discovery method, and connection count.
5. Hub nodes (most connections) are automatically positioned at the center of the layout.

Downstream impact

If a group goes unhealthy, trace its outbound edges to identify which downstream services depend on it and may be affected.

Upstream dependencies

If a service is degraded, trace inbound edges to identify upstream groups whose health may be the root cause.

List view

Switch to the list view for a tabular representation of all dependencies with source/target, port, protocol, confidence, discovery method (auto_traffic vs manual), and connection counts.

Service health rollup

Health-tracked groups compute a health score from their member hosts. The score drives status classification and can trigger alerts.

Health score calculation

• The health computation loop runs periodically, evaluating each group with a health calculation method configured.
• Member host health scores are aggregated using the configured method (Weighted Average, Worst Case, or Majority).
• Alert penalties may further reduce the score if members have active alerts.
• The computed score is stored as a health score metric for historical tracking.

Health history

1. Click a health-tracked group to open its detail view, then select the Health History tab.
2. Returns a time series of health scores, useful for identifying degradation trends.

Status thresholds

Integration with patch management (service groups)

Service groups enable coordinated multi-server patching that respects service dependencies and ordering constraints.

Service group structure

A service group contains ordered tiers. Each tier references a host group and controls:

Managing service groups

1. Navigate to Host Groups > Service Groups tab.
2. Click Create Service Group.
3. Add tiers, each referencing an existing host group.
4. Set the tier order, concurrency, and success gates.
5. When a patch deployment references this service group, hosts are patched tier-by-tier in the defined order.

Visual service map (React Flow)

The Dependencies tab renders an interactive topology visualization using React Flow v12.

Graph features

• Nodes: Represent host groups (colored by health status) or standalone hosts. All host groups in the organization appear as nodes, not just health-tracked ones.
• Edges: Represent dependencies between nodes. Style varies by confidence level.
• Interactive selection: Click any node or edge to open a detail panel with full metadata.
• Spiral layout: Hub nodes (most connections) are centered, leaf nodes are positioned outward.
• Summary stats bar: Shows total node count, edge count, and counts of auto-discovered vs manual dependencies.
• List view toggle: Switch between graph and tabular views.
• Create/delete modals: Add or remove dependencies directly from the Dependencies tab.

Auto-discovery pipeline

Network flow data drives automatic dependency discovery without manual configuration.

1. Agents collect TCP/UDP connections every 300 seconds (configurable).
2. Flow data is submitted by agents and stored for analysis.
3. The dependency discovery engine reads flows and resolves destination IPs to known hosts.
4. It builds group-level dependency records by looking up which host groups each source and destination host belongs to.
5. Stale dependencies decay: confidence halved at 24 hours, deleted at 7 days without new traffic.
6. Manual dependencies are never modified by auto-discovery.

Permissions reference

Navigation reference

Troubleshooting