3.3. Topology map

A live visualisation of the dependencies and traffic flow between applications and services.

Overview

The topology map shows the connections between every application you monitor as a directed graph. You can see at a glance which services each application talks to, how much traffic there is, and whether errors are occurring.

Use it in a microservices environment to identify quickly which services are affected when something breaks, or simply to understand the service architecture in normal times. It is connected to the server in real time, so the topology data refreshes on its own.

Open it from the Topology map menu in the left sidebar.

Screen layout

The topology map is made up of the following areas.

Area	Position	Description
Top header	Top of the screen	Page title, full-screen button, connection badge, alerts button, search, category and namespace filters
Map canvas	Centre of the screen	The main area, where service nodes and their connections are drawn
Zoom controls	Bottom left	Zoom in and out buttons, the current zoom level, and a fit-to-screen button
Legend	Bottom left (below the zoom controls)	The legend for node state colours and traffic line thickness
Minimap	Bottom right	A reduced view of the whole topology and the position of the current viewport

Connection badge

The connection badge in the header shows the state of the real-time connection to the server.

• Live (green dot): topology data is updating in real time.
• Connecting… (grey spinner): connecting to the server.
• Disconnected (red dot): reconnection attempts passed their limit and the connection is down. Refresh the page.

If the connection drops, an error bar appears at the top and you can restore the connection with the Reconnect button.

Main features

How to read a service node

Each application appears on the map canvas as a card-shaped node. A node carries the following information.

Item	Position	Description
Service name	Top of the card	The application's name
Status badge	Below the name, left	The current state (OK, Warning, Critical)
Pod count	Below the name, right	How many pods are running
CPU gauge	Bottom of the card	CPU usage (percent)
MEM gauge	Bottom of the card	Memory usage (percent)
Status bar on the left	The vertical bar down the left of the card	Colour-coded by state

The node's border colour and its left-hand status bar represent the service's state.

Colour	State	Meaning
Green	OK	The service is running normally
Yellow	Warning	Needs attention
Red	Critical	Needs immediate action

Nodes in a warning or critical state get a static highlight ring around their border. Nodes whose CPU or memory usage passes 90% are highlighted the same way.

Gauge bar colours also change with usage.

Colour	Threshold	Meaning
Green	Below 75%	Normal
Yellow	75% to 90%	Caution
Red	90% and above	Critical

External service nodes

Services outside your monitoring (an external API or a SaaS, say) are drawn in a reduced form. External service nodes have no CPU or memory gauges — only the service name and a status dot.

Namespace areas

Nodes belonging to the same namespace are grouped inside a dashed border.

• The namespace name and its service count appear at the top of the area.
• Every namespace area uses the same neutral-coloured dashed border.
• Click a namespace name to apply that namespace filter quickly. Click again while it is applied and the filter clears.

Note: a namespace is the unit Kubernetes uses to separate resources logically.

How to read the connections

Traffic between services is drawn as connecting lines with arrows, which carry the following information visually.

Visual element	Meaning
Arrow direction	The direction traffic flows (from the caller to the callee)
Line thickness	Requests per second (RPS). More traffic, thicker line
Line colour	Error rate. Green (normal) / yellow (above 1%) / red (above 5%)
Flow animation	Where there is traffic, a dashed line animates along the connection
Inline figures	Requests per second, p50 latency, and bytes sent and received, shown on the line

The speed of the flow animation follows the volume of traffic — the more requests per second, the faster it flows.

Node and connection tooltips

Hover over a node and a detail tooltip appears, containing the following.

• Service name and status badge
• Namespace
• CPU usage (percent and gauge)
• Memory usage (percent and gauge)
• Pod count
• Inbound and outbound traffic (requests per second)
• Error rate
• Per-inspection status, where present

Hover over a connection and the tooltip gives the detailed figures for it.

• The direction, from calling service to target service
• Requests per second (RPS)
• Error rate
• p50 and p95 latency
• Bytes sent and received per second

Hover highlighting

Hover over a node and only that node, the nodes directly connected to it and the connections between them stay highlighted; everything else dims. This makes a service's upstream and downstream dependencies easy to see.

Hovering over a connection likewise highlights it and the nodes at both ends.

Clicking a node for details

Click a service node on the map and the detail dialog for that application opens, where you can look at metrics, logs, distributed traces and other data.

Double-click a node and the map moves to it at 100% zoom.

Note: clicking an external service node does not open a detail dialog.

Checking alerting nodes

Click the alerts button (the warning icon) in the header to see the services currently in a warning or critical state.

• Where there are alerts, the button carries a red count badge to draw attention.
• With no alerts, a "No alerts" message is shown.
• The list is sorted with critical first, then warning.
• Each entry shows a status colour dot, the service name and the namespace.
• Click an entry and the alert list closes and the map moves to that node.

Tip: when you first open the topology map, if any node is in a critical or warning state, the map focuses on it automatically.

Searching for a service

Type a service name into the search box on the right of the header to find a service quickly. Autocomplete is supported.

1. Type a service name into the search box.
2. Choose the service you want from the dropdown.
3. The map moves to that service's node.

Note: the search list contains only the services currently on screen under the category and namespace filters in effect.

Filtering

Use the filters in the header to narrow which services appear on the map.

• Category filter: shows only services in the chosen categories. Several categories can be selected at once.
• Namespace filter: shows only services in the chosen namespaces. Several namespaces can be selected at once.

With a filter applied, only the matching services and the connections between them appear, and the map layout rearranges itself to fit the screen.

The first selection appears as a chip; further selections are summarised as a +N badge. Clear an individual filter with the X button on its chip, or clear them all with the X button on the dropdown.

Tip: clicking a namespace area's name on the map canvas applies that namespace filter quickly. Click again while it is applied and the filter clears.

Navigating the map

Navigate the map canvas as follows.

Action	How
Pan	Drag an empty area
Zoom	Scroll the mouse wheel (zooms around the cursor)
Double-click an empty area	Moves to 100% zoom centred on the cursor
Double-click a node	Moves to 100% zoom centred on that node

Zoom controls

The zoom controls at the bottom left of the map canvas offer the following.

Button	Function
- (zoom out)	Zooms the map out
Zoom level	Shows the current zoom level as a percentage
+ (zoom in)	Zooms the map in
Fit to screen	Adjusts the view so every service node fits on screen

Zoom ranges from 30% to 150%.

Legend

The legend sits below the zoom controls.

• Node state colours: the colour for each of OK, Warning and Critical
• Connection thickness: light traffic (thin line) and heavy traffic (thick line)

Minimap

The minimap sits at the bottom right of the map canvas and gives a reduced view of the whole topology.

• The rectangular frame marks the area currently on screen (the viewport).
• Click or drag on the minimap and the map moves to that position.
• Each node appears as a dot coloured by its state.
• The area outside the viewport is dimmed, so you can see at a glance which part you are looking at.

Full-screen mode

Click the full-screen button in the header and the topology map fills the screen, letting you take in a complex service structure at once.

Click the button again while in full screen, or press Esc, to return.

Worked examples

Working out the blast radius of a failure

1. When you open the topology map, it focuses automatically on any node in a critical or warning state.
2. Hover over the problem node to see its dependencies. Only the upstream and downstream services directly connected to it stay highlighted.
3. Check the connection colours to see along which paths errors are spreading.
4. Click the problem node and check its metrics, logs and distributed trace data in the detail dialog.

Understanding the service architecture

1. Click the fit-to-screen button to take in the whole topology.
2. Look at the service groups enclosed by namespace areas.
3. Read traffic flow and volume between services from the direction and thickness of the connections.
4. Where you are curious about one service's dependencies, hover over its node and read the highlighting.

Focusing on one namespace

1. Choose the namespace to analyse in the namespace filter, or click the namespace name on the map canvas.
2. Only that namespace's services remain, and the map rearranges itself.
3. Analyse the internal communication patterns between the services and their external dependencies in detail.

Related documents

• Dashboard - overall cluster resource status
• Applications - detailed monitoring of an individual application
• Incidents - detected incidents and their analysis
• Timeline map - the relationship between deployments and incidents