{"canonical_url":"https://engineer.company/glossary/dashboard/","content_markdown":"# Dashboard\n\n## What it is\n\nA dashboard is a page of panels, each one a query turned into a picture, laid\nout so that someone can read the state of a thing without asking anyone. The\nsame idea goes by other names: Prometheus's own documentation calls them\nconsoles, and a team might say \"the board\" or \"the screen on the wall\". Grafana\ndescribes its dashboards as one or more panels arranged in rows or tabs that\n\"provide an at-a-glance view of related information\", and each panel as the\ncontainer for a visualisation built from a query against a data source. That\nsource can be a time-series store, a SQL database, an API or a plain CSV file.\n\nThe panels come in a handful of shapes. A line over time shows a trend: memory\nin use across the last day, requests per minute across the last week. A single\nlarge number answers \"how much right now\". A gauge shows a value against a\nknown range, such as disk used out of disk available. A table lists things\nthat need attention, one row each. A status light says up or down. Most good\ndashboards use two or three of these shapes, picked for the question, and very\nfew panels of each.\n\nOn a web page those panels are ordinary interface parts, and the W3C's ARIA\nAuthoring Practices Guide describes how each should behave for people using a\nkeyboard or a screen reader. Tabs switch between views and show one panel of\ncontent at a time. A table is a static grid of rows and cells; a grid is the\ninteractive version, moved through with the arrow keys. A meter is the pattern\nfor a value inside a defined range, and an alert is a short, important message\nthat draws attention without interrupting what the reader is doing. A gauge\nwritten to the meter pattern carries its range and its reading as text:\n\n```html\n\u003cdiv\n    role=\"meter\"\n    aria-label=\"Disk used\"\n    aria-valuemin=\"0\"\n    aria-valuemax=\"100\"\n    aria-valuenow=\"72\"\n    aria-valuetext=\"72% (31 GB free)\"\n\u003e\u003c/div\u003e\n```\n\nA dashboard waits to be looked at, and that sets it apart from its neighbours.\nAn alert comes to a person when something crosses a line. A report is a\nsnapshot, produced when asked for and then fixed. A dashboard is live, and it\nis worth exactly as much as the attention someone gives it.\n\n## Why it matters\n\nA dashboard gives a team one shared reading of the state of things. Without\nit, the same question gets answered by whoever has a terminal open, in their\nown way, and two people in an incident can spend minutes discovering that they\nwere looking at different numbers. With it, the conversation starts from the\nsame picture, and people who would never write a query can see how the system\nis doing.\n\nIt also shortens the walk from \"something is wrong\" to \"this part is wrong\".\nThe Prometheus guidance on consoles recommends building one dashboard per\nservice, each showing the latency and errors of the services it talks to, so\nthat someone can start at the top of the tree and follow the bad numbers down\nto the service causing them. A dashboard laid out along the structure of the\nsystem turns that walk into a few clicks.\n\nIt has a cost too. Every panel runs its query on every refresh, on every screen\nthat has it open. Every panel is also a claim that someone keeps true when the\nservice behind it is renamed, split or retired. A dashboard with forty panels\nhas forty of those claims, and most of them go unread.\n\n## What it looks like in practice\n\nIn an earlier role, the founder built an organisation's\n[first observability dashboard](/portfolio/built-the-companys-first-observability-dashboard/), which\nstarted by collecting CPU, RAM, disk and temperature from remote Linux servers\nover SSH, Docker containers included. A second version moved to Grafana and\nPrometheus, and the panels on it were chosen with the DevOps and engineering\nteams: CPU utilisation, memory usage, service uptime and API latency. It ran on\nthe large TV in the office, where anyone walking past could see the state of\nproduction, and alerting on threshold breaches went alongside it. Teams\ndetected and resolved incidents 40% faster.\n\nA later\n[monitoring dashboard on Grafana and Prometheus](/portfolio/cut-operational-risk-with-grafana-and-prometheus/)\nnames the part that mattered most: metrics that meant something, rather than\nvanity numbers that look busy and tell nothing. Alert thresholds were then set\non those metrics and surfaced where the team would see them while there was\nstill time to act. The team moved from firefighting to catching problems while\nthey were small.\n\nOn a host with 464 MB of memory, the same kind of question gets a different\nanswer.\n[Seven read-only roles render the live host into Markdown reports](/portfolio/built-7-read-only-host-reporting-roles/)\non demand, covering access, traffic, metrics, security and more, because a\nmetrics stack would not fit and would not repay its cost if it did. Their\ngoverning rule suits any dashboard just as well: print only numbers the run\nmeasured, since a report that fills a gap with a plausible figure is worse than\none that leaves it blank.\n\n## Where it goes wrong\n\nA dashboard can look healthy and still mislead, and most of the ways it does\ncome from what was put on it.\n\n**Every available number goes on the screen.** It is tempting to show\neverything a system can measure. The Prometheus guidance warns that this\nproduces consoles so dense that even an expert struggles to read them, and\nsuggests at most five graphs on a console and at most five lines on each graph.\nStart from the likely failures and show what tells them apart.\n\n**Vanity numbers take the space.** A total of requests served since launch\nonly ever rises, and a count of registered users says little about today. They\nlook busy and change no decision. Give the panel to a number that would make\nsomeone act if it moved.\n\n**One dashboard tries to serve everyone.** What an on-call engineer needs at\nnight, which is what is broken, differs from what a developer needs while\nbuilding a feature, such as how many people hit a particular edge case. The\nPrometheus guidance says plainly that one set of consoles rarely serves both,\nand two sets work better.\n\n**Colour carries the whole message.** A row of red and green dots tells nothing\nto someone who cannot tell red from green, nor on a monochrome screen. WCAG 2.2\nsuccess criterion 1.4.1, at Level A, says colour must not be the only visual way\nof conveying information; add a word, a shape or a position. Under 1.4.11, at\nLevel AA, the parts of a chart needed to understand it need a contrast ratio of\n3:1 or more against the colours beside them, which thin pastel series on a white\nbackground often miss.\n\n**A gauge implies a ceiling that is absent.** The APG meter pattern is for a\nvalue within a defined range, and its own note warns against using one for a\nvalue with no meaningful maximum, or for progress. Requests per second has no\nnatural top, so a dial for it invents one and makes an ordinary busy hour look\nlike an emergency. Use a line over time instead.\n\n**The screen waits for a viewer who has gone home.** A dashboard on an office\nTV at three in the morning is watched by nobody. Problems that need a person\nbelong in alerts, and the Prometheus alerting guidance recommends alerting on\nthe symptoms users feel and linking each alert to the console that helps find\nthe component at fault. The alert summons someone, and the dashboard is where\nthey land.\n\n## Further reading\n\n- [Grafana: Dashboards](https://grafana.com/docs/grafana/latest/visualizations/dashboards/)\n  — panels, rows, data sources, queries and transformations, as one widely used\n  tool defines them.\n- [WAI-ARIA Authoring Practices Guide: Patterns](https://www.w3.org/WAI/ARIA/apg/patterns/)\n  — how tabs, tables, grids, meters and alerts should behave for keyboard and\n  screen-reader users.\n- [Prometheus: Consoles and dashboards](https://prometheus.io/docs/practices/consoles/)\n  — the five-graph, five-line guideline and the case for one audience per set.\n- [Prometheus: Alerting](https://prometheus.io/docs/practices/alerting/) —\n  alerting on symptoms, and linking each alert to the console that explains it.\n- [Understanding WCAG 2.2: Use of Color](https://www.w3.org/WAI/WCAG22/Understanding/use-of-color.html)\n  — why colour alone cannot carry a status, with examples.\n- [Understanding WCAG 2.2: Non-text Contrast](https://www.w3.org/WAI/WCAG22/Understanding/non-text-contrast.html)\n  — the 3:1 contrast that chart lines and interface parts need.\n\n\u003chttps://engineer.company/glossary/dashboard/\u003e\n","date_modified":"2026-11-08T02:22:23Z","description":"How a dashboard is put together from panels and queries, who it should serve, how to keep it readable, and the ways one misleads the people watching it.","formats":{"gemtext":"https://engineer.company/glossary/dashboard/index.gmi","gopher":"https://engineer.company/glossary/dashboard/gophermap","html":"https://engineer.company/glossary/dashboard/","json":"https://engineer.company/glossary/dashboard/index.json","markdown":"https://engineer.company/glossary/dashboard/index.md","pdf":"https://engineer.company/glossary/dashboard/index.pdf"},"kind":"page","language":"en","title":"What is a dashboard?","url":"https://engineer.company/glossary/dashboard/","version":1}