DEX Overview

1. Introduction: What Is DEX?

• DEX measures the health of a device. DEX shows the experience of the user, not only raw data.
• The agent measures the device continuously on Windows, macOS, and Linux. The agent uses few resources.
• DEX does these steps: find a problem, calculate a score, send an alert, repair the problem, make sure that the repair was good.
• DEX is different from other tools:
  • DEX calculates one health score from many signals. It does not only show raw counters.
  • DEX learns the usual values for each device. It sets its own limits.
  • DEX can repair problems. It measures the result of each repair.
  • DEX records the opinion of the user together with the measured data.

2. How DEX Operates

• TD DEX functions on the same core platform as TD ITAM.  This means the Agent and Server components are the same, and feature activation of the two platforms is based on the installed license file.
• The agent measures many health signals in the background.
• Each signal has its own measurement interval. Fast signals have short intervals. Slow signals have long intervals.
• The agent keeps the measurements on the device. The data is kept when the device starts again.
• The agent sends only important data to the server: alerts, snapshots, and summaries. It does not send all raw data.
• The server collects the data from all devices. The server shows trends and comparisons.
• The agent also measures its own resource usage. Two metrics show this.

3. What DEX Measures

• DEX measures more than 40 health metrics in 15 categories.
• The categories include hardware health, resource pressure, application behavior, network quality, and the opinion of the user.
• Some metrics are conditional. Battery metrics operate only on devices with a battery. Data-volume metrics operate only when the device has a data volume on a different disk.

Metric reference table

4. Health Scores

• The Health Score is one number from 1 to 100. DEX calculates it from many signals.
  • Component health (disk, battery, sleep): a high value is good.
  • Resource pressure (CPU, memory, disk, network): a high value is bad. DEX inverts these values in the score.
  • Long-term problems have more weight than short peaks. For some signals, DEX uses the worst recent value. Thus, short but serious problems stay visible.
• Users see three health bands: Healthy, Fair, and At Risk.
• A score is better than a raw value. For example, "Memory Pressure" shows a problem before "% RAM used" shows it.

5. Alerts: Warning Levels and Critical Levels

• Each metric has two limits. The warning limit shows the first problem. The critical limit shows an urgent problem.
• Show a small table with examples. For example: disk usage, warning at 90%, critical at 97%.
• DEX prevents alerts that are not necessary:
  • The trigger limit and the reset limit are different. This prevents alerts that come and go quickly.
  • A condition must continue for more than one interval before DEX sends an alert.
  • Some metrics use an average across time, not one value.

6. Limits That DEX Learns

• Some values are only bad when you compare them with the usual values of the same device. Examples: boot time, wake delay, and background process count.
• DEX learns the usual value for each device automatically. DEX sets the limit at the usual value plus a margin.
• DEX keeps the learned values and adjusts them with time. Manual adjustment for each device is not necessary.

7. Measurements That Show the Experience of the User

• Measurements that DEX takes when the user is active have more weight.
• Measurements from the night and from idle time have less weight. Thus, the scores show the experience during work.
• Call metrics operate only when a call is active.

8. What the User Sees

• The health panel of the agent shows the full health score, the component health, and the resource pressure.
• The user can send feedback: free reports, answers to questions from DEX, and reports about hardware problems.
• The user can stop DEX reports.

9. Automatic Repair

• When a metric goes above its limit, DEX can start a repair.
• The repair sequence:
  1. DEX finds the problem and selects the applicable repairs.
  2. DEX shows the repairs to the user, or does the repairs without a prompt. The policy controls this.
  3. The repairs run.
  4. DEX measures the metric again. DEX also asks the user if the device is better.
• DEX calculates a score for each repair. Thus, the system learns which repairs are good.

10. Reports for the Full Fleet

• The server keeps: a log of alerts, the current data for each device, the daily trends, and the totals for the fleet.
• Administrators can find the devices that are different from the fleet.
• Examples of reports: old batteries, disks that will fail, devices that boot slowly, and network problems.
• The server data can start tickets, notifications, and automatic actions.

11. Privacy and Resource Usage

• DEX does not read files (names or content). DEX does not record web usage (ITAM can track URL usage). Process metrics show only resource usage.
• DEX sends summaries and alerts. DEX does not send continuous raw data.
• The user can stop DEX reports.
• The DEX/ITAM agent is extremely light weight in size and resource use.  The self-measurement metrics of the agent show this, and any issue would be a trigger.

12. Platforms

• Fully supported on Windows, MacOS, and Linux (Debian and RPM).
• Some metrics are not available on all platforms.  See the DEX Platform Support for full details.

13. First Steps and Configuration

• DEX operates with the default configuration. Adjustment is not necessary.
• Administrators can change:
  • the metrics that are on or off
  • the intervals
  • the limits
  • the repair policy.
• Server setup and agent deployment therefore require minimal effort, and if you are already an ITAM customer almost no additional effort will be needed.  See the Setup and Deployment guide for more details.