Exploring Visualization Types

Introduction: Different Questions Need Different Visualizations

In the previous lesson, you created your first Grafana dashboard with two panels: a time series chart showing CPU usage trends over time and a stat visualization displaying the average memory usage as a single number. These two visualization types work great for certain monitoring questions — time series helps you see how metrics change over time, while stat gives you a quick summary value at a glance. However, monitoring systems involves asking many different types of questions about your data. Sometimes, you need to compare average values across multiple hosts to identify which one is consuming the most resources. Other times, you need to see if a current value has crossed into a dangerous threshold zone. You might want to understand how disk space is distributed across different mount points, or you might need to see raw, detailed records for troubleshooting. Each of these questions requires a different visualization type, and each visualization type expects your query to return data in a specific structure. In this lesson, you'll learn four additional visualization types that complement the time series and stat panels you already know: bar charts for comparing aggregated values, gauges for showing current values with color-coded thresholds, pie charts for displaying proportional distributions, and tables for presenting detailed multi-column data. By the end of this lesson, you'll be able to choose the right visualization for each monitoring question you need to answer.

Bar Charts: Comparing Averages Across Multiple Items

Join the 1M+ learners on CodeSignal

Be a part of our community of 1M+ users who develop and demonstrate their skills on CodeSignal

Pie Charts: Distribution Across Categories

When you need to understand how a total is divided among different parts, pie charts provide an intuitive visualization. For instance, you might want to see how disk space is distributed across different mount points on a single host. While a bar chart could show this as well, a pie chart emphasizes the proportional relationship — how each part contributes to the whole.

Create a new panel with this query:

SQL

SELECT
  mount AS metric,
  AVG(used_pct) AS value
FROM public.metrics_disk
WHERE host = 'host-a'
  AND $__timeFilter(ts)
GROUP BY mount
ORDER BY value DESC;

SELECT
  mount AS metric,
  AVG(used_pct) AS value
FROM public.metrics_disk
WHERE host = 'host-a'
  AND $__timeFilter(ts)
GROUP BY mount
ORDER BY value DESC;

This query structure looks similar to the bar chart query because both are creating categories for comparison. You're grouping by mount to create one result row per mount point and calculating the average disk usage percentage for each. The difference is in how you'll visualize the results.

The metric column (mount point names) determines what slices appear in your pie chart, and the value column (average usage percentage) determines the size of each slice. By filtering to a single host with WHERE host = 'host-a', you ensure the pie chart shows the complete picture for one machine rather than mixing data from multiple hosts.

After running the query, switch the visualization type to Pie chart. Then configure the panel so Grafana uses your columns correctly:

Under Value options → Show, choose All values so each mount point becomes its own slice.
Under Value options → Fields, select value to use the numeric column for slice size.
Under Pie chart → Labels, select Name and Value to display the mount point names and values as slice labels.

With these settings applied, you’ll see a circular chart divided into colored slices, each representing one mount point. The size of each slice corresponds to that mount point’s disk usage percentage. Larger slices immediately stand out, helping you identify which mount points are consuming the most space.

Pie charts work best when showing parts of a whole, particularly when you have a small number of categories (typically 5–8 slices). For disk monitoring, this works well because most systems have only a handful of mount points. Larger categories appear more prominently, making it easy to visually compare usage across disks.

Tables: Detailed Multi-Column Raw Data

All the visualizations you've learned so far focus on presenting data graphically, but sometimes you need to see the actual raw records with all their details. This is especially useful for troubleshooting or when you need to drill into specific values. Tables give you this detailed view.

Add one more panel with this query:

SQL

SELECT
  ts AS "Time",
  host AS "Host",
  used_mb AS "Memory Used (MB)"
FROM public.metrics_mem
WHERE $__timeFilter(ts)
ORDER BY ts DESC, used_mb DESC
LIMIT 100;

SELECT
  ts AS "Time",
  host AS "Host",
  used_mb AS "Memory Used (MB)"
FROM public.metrics_mem
WHERE $__timeFilter(ts)
ORDER BY ts DESC, used_mb DESC
LIMIT 100;

This query returns multiple columns of data, and each column will become a column in your table. Notice the use of column aliases with quotes like AS "Time" and AS "Host". These aliases become the column headers in your table, so you're writing them in a readable format with proper capitalization and spacing.

The query selects three pieces of information for each memory measurement: the timestamp, which host it came from, and how much memory was used in megabytes. By ordering with ORDER BY ts DESC, used_mb DESC, you ensure the most recent records appear first, and within each timestamp, the highest memory consumers appear at the top. The LIMIT 100 clause keeps your table from becoming overwhelming by restricting it to the 100 most relevant rows.

Change the visualization type to Table. Unlike the graphical visualizations you've been using, the table will display your data in rows and columns, similar to a spreadsheet. Each row represents one measurement record, and each column shows a different attribute of that measurement.

Tables are particularly valuable when you need to see exact values rather than trends or comparisons. For example, if you notice a memory spike in a time series chart, you might add a table panel to see exactly which hosts experienced high memory usage and at what specific times. The tabular format also makes it easy to identify patterns across multiple dimensions simultaneously — you can see which hosts consistently appear among the top memory consumers or notice if certain times of day show higher usage.

Summary and Practice Preview

You've now learned six different visualization types that work together to create comprehensive monitoring dashboards. Time series charts track how metrics change over time, while stat panels display single summary values at a glance. Bar charts let you compare aggregated values across categories like hosts or services. Gauges show current values with color-coded thresholds for immediate status assessment. Pie charts reveal proportional distributions when you need to understand how a total is divided. Tables present detailed multi-column data for troubleshooting and detailed analysis.

Each visualization type requires a specific query structure. Time series queries use $__time(ts) to identify the time dimension and return metric-value pairs over time. Bar and pie charts use GROUP BY with aggregation to create categories for comparison, without the time dimension. Gauges return a single latest value by filtering to one item and using ORDER BY ts DESC LIMIT 1. Tables return multiple columns with readable aliases to create spreadsheet-like views of raw data. Despite these differences, all queries still use $__timeFilter(ts) to respect your dashboard's time range controls, ensuring consistency across panels.

In the upcoming practice exercises, you'll add all four of these new visualization types to your dashboard: a bar chart comparing CPU usage across hosts, a gauge showing current disk usage with thresholds, a pie chart displaying disk space distribution across mount points, and a table listing top memory consumers. By combining these different visualization types on a single dashboard, you'll be able to answer multiple monitoring questions at once — seeing both the big-picture trends and the detailed specifics.

Looking ahead to the next unit, you'll learn about template variables, which transform these static panels into interactive dashboards. Instead of hardcoding values like host = 'host-a' in your queries, you'll create dropdown selectors that let you dynamically filter data. This means you could have a single gauge panel that shows disk usage for whichever host you select from a dropdown, rather than creating separate gauges for each host. Variables will make your dashboards much more flexible and powerful.

Introduction: Different Questions Need Different Visualizations

Bar Charts: Comparing Averages Across Multiple Items

Join the 1M+ learners on CodeSignal

Query: Return a Single “Current” Value

Switch to Gauge + Configure Value Options and Range

Add Thresholds for Health/Warn/Critical

When to Use Gauges

Pie Charts: Distribution Across Categories

Tables: Detailed Multi-Column Raw Data

Summary and Practice Preview