An interface to manage 10,000 PCs: seven decisions from a designer's practice

Administrators work on 13-inch laptops with browser panels open, so the useful window height is about 630 pixels. Mockups on large monitors hide that reality. Hence the project's first rule: every vertical pixel is worth more than a horizontal one.

Space first, then complex interactions, then error prevention.

• A compact navigation rail instead of a full sidebar.
• A collapsible filter panel so the table gets maximum area.
• A select menu instead of tabs: new sections do not break the layout.
• An 8 px micro-grid as a functional rather than aesthetic tool.
• Layered drawers instead of modal windows: context stays, previous data remains visible.
• Master-Detail with a timeline that shows task density over time.
• Splitting collisions into logical (schedule conflicts) and resource (network overload) with indicators right in table rows.

In a system where one action touches thousands of machines the interface must make errors hard. Creating a task requires a mandatory 'On failure' block: what to do if some machines do not respond. A load heat map shows when a task exceeds network capacity: a red line, not a paragraph in documentation.

The author sets two metrics: time on task, reduced by mask search and bulk operations, and error rate, reduced by conflict validation and the absence of blind spots. Both are measurable before and after and independent of taste.

An enterprise interface is designed from the real screen and the cost of an error, not from a pretty mockup. Rail, drawers, timeline and a mandatory failure scenario work in website admin panels and client areas too.