Library Look up only what you need, then get back to the task.

The System Feels Slow

Use this when the machine feels sluggish and you need to decide whether the pressure is CPU, memory, disk, startup load, or one runaway process.

Troubleshooting playbook

The System Feels Slow

The system is slow, laggy, or less responsive than usual.

Processes, Performance, and Services medium severity both
Symptom

The system is slow, laggy, or less responsive than usual.

Meaning of the symptom

“Slow” is only the user experience. The real cause still needs to be narrowed.

Safe sequence

  1. Check CPU pressure.
  2. Check memory pressure.
  3. Check disk activity and wait.
  4. Identify top consumers.
  5. Decide whether the issue is one process, many background tasks, or a startup/service pattern.

Common branch decisions

  • High CPU with one process often points to one runaway workload.
  • High memory with lots of swap/paging points to memory pressure.
  • High disk wait with normal CPU often means storage or background service contention.
  • If the slowdown is mostly at boot or login, examine startup and service load next.
First checks
  • Check CPU, memory, and disk pressure separately.
  • List the top processes by current resource usage.
  • Notice whether the slowdown is global or tied to one app or task.
Likely causes
  • One runaway process or browser workload.
  • Low available memory causing heavy paging or swapping.
  • Disk pressure, especially from background scans, updates, or indexing.
  • Too many startup or background tasks competing at once.
Do not do yet
  • Do not reboot before gathering any evidence.
  • Do not kill processes unless you know what they are.
  • Do not assume memory is the problem if disk or CPU is saturated.
Verify fix
  • Resource pressure drops after the intended action.
  • Responsiveness returns in the actual workload, not just in one metric.
  • You can explain what subsystem was really under stress.
You own this when
  • You can inspect the main performance signals before acting.
  • You can choose a fix that matches the real bottleneck.
  • You can verify improvement with the same workload that felt slow.
Reflect before you leave
  • What signal helped you separate CPU, memory, and disk pressure?
  • What would have been misleading about guessing based on one high number alone?

Revisit this playbook in about 1, 7, 21 days.