6.1 Host Map and Infrastructure List
Key Takeaways
- Host Map is Datadog's visual infrastructure explorer for hosts, containers, pods, and clusters — default Fill by metric is CPU utilization.
- Group by arranges tiles into sections by tag keys such as env, service, or availability_zone; Size by adds a second metric dimension via tile size.
- Infrastructure List is the tabular counterpart to Host Map, showing the same inventory with sortable columns and tag-based filtering.
- Both views are read-only analysis tools — they do not modify host tags, Agent config, or monitor definitions.
- Clicking a host tile opens a side panel with key metrics, processes, and links to related dashboards and logs.
Why Infrastructure Views Matter on the Exam
The Data Visualization & Utilization domain expects you to move beyond raw metrics and use Datadog's built-in views to answer "what is unhealthy, where, and in what context?" Two views dominate that workflow: Host Map and Infrastructure List. Both consume the same underlying host and container inventory the Agent reports, but they present it differently — one as a spatial heat map, the other as a sortable table. The Fundamentals exam frequently tests what each control does, what the defaults are, and what these views cannot do.
Host Map: Visual Infrastructure Exploration
Host Map renders infrastructure entities — physical hosts, virtual machines, containers, pods, and clusters — as tiles on a canvas. Each tile represents one entity (or a grouped cluster of entities, depending on zoom level). The map is designed for pattern recognition: you can spot a single red-hot host in a green fleet, see whether an outage is isolated to one availability zone, or compare container density across Kubernetes nodes.
Three controls define how the map communicates:
| Control | Purpose | Exam default |
|---|---|---|
| Fill by | Metric that determines tile color intensity | CPU utilization (percentage) |
| Size by | Optional second metric that determines tile size | None (all tiles equal size) |
| Group by | Tag key that arranges tiles into labeled sections | None (flat layout) |
Fill by is the most-tested setting. By default, Host Map colors each host or container by CPU utilization, so saturated processors appear as warm or red tiles while idle systems stay cool. You can change Fill by to memory usage, disk I/O, or any supported metric — the exam expects you to know that Fill by controls color, not size or grouping.
Size by adds a second visual dimension. A common pattern is Fill by CPU and Size by memory: a large, red tile might indicate a host with both high CPU and high memory pressure. Size by is optional; when disabled, all tiles are the same size regardless of metric values.
Group by spatially arranges tiles into sections based on a tag key or Datadog attribute. Grouping by env separates production from staging; grouping by service clusters all web servers together; grouping by availability_zone reveals whether a problem is regional. Group by is a display control — it does not mutate underlying host tags, create application keys, or delete hosts that fail to match.
Filtering and Scope
Above the map, tag filters narrow the visible population. A filter like env:prod AND service:api shows only production API hosts. This is the same tag-scoping model used in Metrics Explorer's from box and dashboard template variables. If you see an empty map, the first check is whether your filter is too restrictive — not whether the Agent is broken.
Host Map also supports different views depending on your infrastructure: hosts, containers, pods, tasks, or clusters. In Kubernetes environments, switching to the pod view lets you compare container-level metrics without leaving the map. The exam may describe a scenario with containers or pods and expect you to recognize that Host Map covers all of these entity types, not just bare-metal hosts.
Infrastructure List: The Tabular View
Infrastructure List (reachable from Infrastructure > Host List or as a toggle from Host Map) shows the same inventory in a table with sortable columns: hostname, platform, Agent version, CPU, memory, disk, and configured tags. Where Host Map excels at visual hotspot detection, Infrastructure List excels at fleet-wide comparison — sorting by CPU to find the top ten consumers, filtering by role:database, or checking Agent version consistency across a region.
Both views share the same data source and tag filters. A filter applied in Host Map carries over to Infrastructure List and vice versa. Clicking a row in Infrastructure List opens the same host side panel available from Host Map tiles.
Host Side Panel and Drill-Down
Clicking any host or container tile opens a side panel with a summary of key metrics, running processes, installed integrations, and links to related dashboards, logs, and traces. This panel is the bridge between infrastructure visualization and deeper investigation — the exam may ask what happens when you click a tile (you drill into detail, you do not edit Agent configuration or rotate API keys).
Common Exam Traps
Several wrong answers appear repeatedly:
- Host Map does not compile integrations, rotate API keys, or store log archives. It is purely a visualization and exploration tool.
- Group by does not delete hosts or change their tags — it only rearranges the display.
- Fill by defaults to CPU, not disk mount count, monitor count, or API key age.
- Host Map and Infrastructure List are not required for the Agent to function; they consume data the Agent already sends.
Practical Workflow Example
Imagine on-call receives a latency alert. You open Host Map, set Group by to service, Fill by to CPU, and filter to env:prod. Three API hosts glow red while the rest stay green — the problem is likely localized to those hosts, not a fleet-wide event. You click one tile, confirm high CPU in the side panel, and pivot to a service dashboard. That sequence — map for scope, side panel for detail, dashboard for correlation — is exactly the investigation pattern Datadog teaches and the exam rewards.
By default, what metric does Host Map use to color each host or container tile?
On Host Map, what does the Group by control do?
What is the main purpose of Datadog's Host Map?