Insight: Experience

Updated

Insight: Experience

Overview

Use this page to monitor real-user experience (UX) scores for a machine group's session hosts. Insight Experience turns session telemetry, application launch time, input delay, and host CPU/memory pressure, into a score for each application, compared against that application's own 7-day rolling baseline rather than a fixed target. That tells you whether today's performance is worse than normal, instead of guessing from a one-off complaint.

This page depends on machine groups and session hosts being configured: hosts are registered on Configuration: Insight: Machines and grouped on Insight: Machine Groups, while add-ins on those hosts generate the launch and input-delay measurements the score is built from. The scores and host-pressure state shown here feed the Insight: Dashboard view and are the evidence you'd use to raise an alert or open a ticket about an application or host.

On this page you select the machine group to inspect, read four summary tiles (UX score, launch time, input delay, and hosts under pressure), and drill into a per-application table and a hosts panel, with automatic or manual refresh to keep the figures current.

Key concepts

  • Machine group - A set of session hosts and the applications running on them that are scored together as one unit on this page. (related: Insight: Machine Groups)
  • UX score - A single composite score summarizing real users' experience on a machine group's session hosts, built from launch time, input delay, and host pressure. (related: Insight: Dashboard)
  • 7-day baseline - The rolling average for an application's score, launch time, or input delay over the trailing 7 days, used as the normal-performance reference point for the current window.
  • Hosts under pressure - Session hosts whose CPU or memory usage is high enough to risk degrading the sessions running on them. (related: Configuration: Insight: Machines)

When to use it

  • You want a quick read on whether real users on a machine group's session hosts are currently experiencing degraded launch times, input delay, or host pressure.
  • You need to compare an application's current performance against its own 7-day baseline rather than a fixed target.
  • You are troubleshooting a complaint about slow application launches or sluggish input and want to confirm it with measured data instead of anecdote.
  • You want to check whether any session host in a group is under CPU or memory pressure before it affects more sessions.
  • You are validating that a recent change, such as a new image or a capacity change, improved or didn't regress real-user experience.

Before you begin

  • A machine group exists for the hosts you want to check; create or review one on Insight: Machine Groups.
  • The session hosts in that machine group are registered and configured on Configuration: Insight: Machines.
  • The session hosts have the add-in that captures launch and input-delay telemetry installed, configured on Configuration: LoadGen Agents: Session Host Add-ins.
  • Real user or test sessions have launched applications on those hosts within the time window you plan to review; otherwise the tiles show a dash and the table shows no rows.

Common workflows

Check whether real users are experiencing degraded performance right now

Persona: Load-test operator or IT admin monitoring production session hosts

  1. Select the machine group you want to check from the machine group selector.
    • Expect: The KPI tiles, Applications table, and Hosts panel reload scoped to that group, and the scope line below the toolbar updates to show the group name, active session count, and time window.
  2. Confirm Auto is checked and review the refresh interval next to it.
    • Expect: The Updated Xs ago label keeps advancing on its own, so the tiles stay current without further action.
  3. Read the UX score, Launch p95, all apps, Input delay p95, and Hosts under pressure tiles.
    • Expect: Each tile shows a numeric value with its 7-day baseline underneath when launches occurred in the window, or a dash when none did.

Verify: The UX score tile shows a number instead of a dash and the Applications table lists at least one application row instead of the empty-state message, confirming the page is scoring real launches for the selected group.

Investigate a host under pressure

Persona: IT admin responsible for session host capacity

  1. Check the Hosts under pressure tile for a nonzero numerator, for example 1 / 3 instead of 0 / 3.
    • Expect: A nonzero numerator means at least one host in the group is under CPU or memory pressure instead of being within limits.
  2. Open the Hosts panel below the Applications table.
    • Expect: The panel lists the session hosts for the selected machine group so you can identify which one is under pressure.

Verify: You can name the specific host that is under pressure, so you know which session host to investigate in Configuration: Insight: Machines.

Steps

Step 1

Select Insight in the left navigation, then open Experience.

Expected result: The page loads for the LoadGen Demo machine group with the UX score, Launch p95, Input delay p95, and Hosts under pressure tiles, plus the Applications and Hosts panels below. Because no application launched in the last 5 minutes, the UX score and Launch p95 tiles show a dash and the Applications table reads No application launched in this window.

This is the starting point for checking whether real-user experience on a machine group's session hosts matches expectations before drilling into a specific application or host.

  • Machine group selector defaults to LoadGen Demo; switch it to check another group, such as svc.lg.godwit.
  • Auto is checked with the interval set to 1m, so the tiles keep refreshing without manual action.
  • Refresh forces an immediate reload outside the auto-refresh cycle. Insight: Experience - step 1

Field reference

Field Purpose Format / Validation Required Default
Machine group selector Chooses which machine group's session hosts, applications, and scores this page displays. Switching groups reloads the tiles, the Applications table, and the Hosts panel for the newly selected group. dropdown, single selection No LoadGen Demo
Auto Turns automatic refreshing on or off. Leave it checked to keep tiles and the Applications table current without manual action; uncheck it when you want to inspect a fixed snapshot without the figures changing while you read them. checkbox No checked
Refresh interval Sets how often the page re-pulls data while Auto is checked. A shorter interval surfaces new launches and host-pressure changes sooner at the cost of more frequent requests; a longer interval reduces load but delays when you notice a regression. dropdown No 1m

Column reference

Column What it shows How to read it
UX score Overall experience score combining launch time, input delay, and host pressure for all applications in the selected machine group. Compare the current number against the 7-day baseline shown beneath it; a dash means no application launched in the current window, not a fault.
Launch p95, all apps The 95th percentile application launch time across all applications in the machine group for the current window. A value higher than the baseline means launches are slower than this application set's normal 7-day pattern.
Input delay p95 The 95th percentile measured delay between a user's input and the application's response, across the machine group's sessions in the window. Higher milliseconds than the baseline indicates sluggish input response for real users on these hosts.
Hosts under pressure Count of session hosts in the group whose CPU or memory is under strain, out of the group's total host count. A nonzero numerator means at least one host is straining its CPU or memory and may be degrading the sessions it hosts.
APPLICATION Identifies the application whose launches were scored in the current window. Each row is one application that had at least one launch in the window; the table is empty when none did.
SCORE The application's UX score for the window. A lower score than this application's own baseline indicates a degraded real-user experience for that app specifically.
LAUNCH P95 The 95th percentile launch time measured for this application in the window. Compare against the application's own baseline to see whether launches are slower than normal for it specifically, rather than the machine group as a whole.
VS BASELINE Shows how this application's current score compares to its own 7-day baseline. Use it to spot, at a glance, which applications are performing worse than their own recent normal, without reading the raw numbers.
SESSIONS The number of sessions that launched this application within the window. A low session count means the score is based on fewer launches and may be less statistically reliable.
TREND (24 H) Shows how this application's score has moved over the last 24 hours. Use it to see whether a dip in the current window is a one-off or part of a longer slide over the day.
STATE The application's current health state as determined by the scoring engine. Scan this column first to find which applications need attention before reading the detailed score and launch columns.

Controls reference

Refresh

Immediately re-fetches the UX score, KPI tiles, Applications table, and Hosts panel for the selected machine group, outside the normal Auto refresh cycle.

When to use: Use it right after a test run or real user session completes, when you don't want to wait for the next automatic refresh interval.

Side effects: Read-only: updates the displayed figures and the Updated Xs ago timestamp; does not change any stored data.

Recompute baseline now

Recalculates the 7-day rolling baseline used to compare current UX score, launch time, and input delay for applications in the selected machine group.

When to use: Use it after a change makes the last 7 days unrepresentative of normal performance, for example after fixing an environment issue that had skewed scores, so VS BASELINE reflects healthy performance instead of the affected period.

Side effects: Mutating: overwrites the existing baseline values shown under the UX score and Launch p95 tiles and used in the Applications table's VS BASELINE column; comparisons made against the old baseline are lost.

Verify the result

When the page is working you land on Experience with LoadGen Demo, or your selected machine group, pre-selected, four KPI tiles, and the Applications and Hosts panels below. If sessions have run and applications launched within the current window, the UX score, Launch p95, and the Applications table populate with numeric values and app rows instead of dashes and the empty-state message. The Updated Xs ago label in the toolbar keeps advancing with Auto refresh, or resets after you click Refresh, confirming the page is pulling live data rather than being stuck on stale figures. If scores remain dashed or the table stays empty long after a known run on hosts in that group, the first thing to check is whether you selected the correct machine group and whether the time window actually overlaps the run.

Troubleshooting

  • Symptom: The UX score and Launch p95 tiles show a dash. Fix: Confirm at least one application launched on a session host in the selected machine group within the current window; a dash is expected when there is no launch to score, not a fault.
  • Symptom: The Applications table always reads No application launched in this window. Fix: Verify you selected the machine group that actually ran the test or real sessions, and that the session host add-ins are installed and reporting for that group's machines.
  • Symptom: Hosts under pressure stays at 0 / 3 even though a host is clearly overloaded. Fix: Check that the overloaded host is actually a member of the selected machine group in Configuration: Insight: Machines; a host outside the group isn't counted here.
  • Symptom: The Updated Xs ago label stops advancing. Fix: Confirm Auto is still checked, then click Refresh to force an update and confirm the page is still reachable.
  • Symptom: VS BASELINE comparisons look wrong after a known environment fix. Fix: Use Recompute baseline now so the 7-day baseline no longer includes the affected period.
Was this article helpful?
0 out of 0 found this helpful