Skip to content

CMDB Data Correctness

Purpose

CMDB Data Correctness goes beyond field completeness to check the integrity of the data in the Clarity Transform CMDB. It counts five kinds of issue for each CI type (configuration item type, such as Application or Device): Provider ID duplicates, name duplicates, stale records, orphaned CIs and dangling references. The on-screen subtitle reads: "Monitor duplicates, stale records, and orphaned CIs across the CMDB."

The page reports issues only. It has no merge, archive or fix action: you correct the records on the CMDB grids.

Where to find it

Discover → Data Quality → CMDB Data Correctness. The page header is CMDB Data Correctness.

What you see

Stale threshold

  • Stale threshold (days): the number of days after which a record counts as stale, 30 by default. An empty value, 0 or anything that is not a number is treated as 30. The value is not remembered: the page opens with the default each time.
  • Refresh: recalculates the cards, chart and table with the threshold you entered. Nothing else on the page uses the threshold.

Cards

Each card has a help tooltip, quoted here.

  • OVERALL CORRECTNESS: a percentage to one decimal place, with the sub-label "Stale > N days", where N is the threshold used. Tooltip: "Overall correctness after accounting for duplicates, stale, and orphaned records."
  • CI TYPES: the number of CI types in the table. Sub-label "Tracked categories". Tooltip: "Number of configuration item categories included in correctness analysis."
  • CRITICAL DUPLICATES: the total of the Provider ID Dupes column. Sub-label "Provider ID dupes". Tooltip: "Provider ID duplicates are a critical data error - the same source record was imported twice."
  • NAME DUPLICATES: the total of the Name Dupes column, with its share of all records ("N% of total"). Tooltip: "Records sharing the same name. May need investigation."

In this build OVERALL CORRECTNESS can show 0.0% while CRITICAL DUPLICATES is 0 and only part of each CI type's records are stale. The screen does not show how the figure is calculated, so read CI Type Breakdown as well as the card.

Correctness by CI Type

A bar chart with the series Provider ID Dupes %, Name Dupes %, Stale % and Dangling Refs % for each CI type. Its tooltip: "Percentage of duplicate, stale and dangling-reference errors for each CI category. Dangling references are counted against record count, so they can exceed 100%." The percentage axis is not capped at 100%.

CI Type Breakdown

A table with one row per CI type. Its tooltip: "Counts per CI category. Click a duplicate count to see the affected records."

  • CI Type
  • Total: the number of records of that CI type.
  • Provider ID Dupes: "Critical: same source record imported more than once. Click count to drill down."
  • Name Dupes: "Records sharing the same name. Click count to drill down."
  • Stale: the number of stale records, followed by their share of the CI type's records as a whole percentage.
  • Orphaned: "CI records that no relationship refers to - this CI is not linked to anything." The count is not a link.
  • Dangling Refs: "The reverse of Orphaned: names that relationships point at but which have no CI record."

In the Provider ID Dupes, Name Dupes and Dangling Refs columns, a count of 0 shows as a tick. A count above 0 is a link that opens a dialog listing the records or names behind it. A row is shaded red when it has Provider ID duplicates or dangling references, and amber when it has name duplicates only.

Messages

  • "Failed to load correctness metrics": a red banner at the top of the page when the figures could not be loaded, for example after Refresh.
  • When there are no CI types to report, the chart and the table each show an empty-state message in place of the data.

How to

Change the stale threshold

  1. Go to Discover → Data Quality → CMDB Data Correctness.
  2. In Stale threshold (days), enter the number of days.
  3. Select Refresh. The OVERALL CORRECTNESS sub-label changes to "Stale > N days" with the new value.

Find duplicate records

  1. On the CMDB Data Correctness page, find the CI type in CI Type Breakdown.
  2. Select its Provider ID Dupes or Name Dupes count. A dialog opens. Its title is the CI type followed by Provider ID Duplicates or Name Duplicates, and its text reads:
    • For Provider ID duplicates: "These records share the same provider_id. This indicates the same source was imported more than once", followed by "a critical data integrity issue."
    • For name duplicates: "These records share the same name. Review and merge or remove duplicates as appropriate."
  3. Select a record's name in the dialog. The CMDB opens in the same browser tab on that CI type's grid, with the name entered in the Search: box. For name duplicates, every record with that name is listed, and records that contain the same text elsewhere can be listed too.
  4. To remove a duplicate, select the record's row by selecting any cell except Name (selecting the Name opens the record's edit form), and select Remove selected. This deletes the record.
  5. Go back to CMDB Data Correctness and select Refresh to see the new counts.

If a device you opened from the dialog is not listed, select Display all scopes...: the Device grid lists only In scope devices until you do. The Application and Database grids hide Out of scope records in the same way. See CMDB.

The duplicates dialog grid remembers its Search: text, sort order and page size in your browser, and the same settings apply to the duplicates dialog of every CI type. If a dialog lists fewer records than its count, clear its Search: box.

Find dangling references

A dangling reference is a name that a relationship points at, but for which no record of that CI type exists. Relationships are matched on name, so anything that joins a record to its relationships skips these, and dependency, risk and lifecycle figures under-report.

  1. On the CMDB Data Correctness page, find the CI type in CI Type Breakdown.
  2. Select its Dangling Refs count. A dialog titled with the CI type and Dangling References lists the names that relationships point at, most-referenced first, so you can fix the ones that affect most relationships first. Its text starts "Relationships point at these names, but no" followed by the CI type and "record exists for them", and ends "Either import the missing records or remove the stale relationships."
  3. For each name, either import or add the missing record, or remove the stale relationships on the Relationship grid of the CMDB. The names in this dialog are not links.

A misspelt name in a relationship shows here as a dangling reference, and the record it should point at can show as orphaned.

An orphaned CI is one that no relationship refers to. To link it, add a record of the Relationship CI type on the CMDB screen, or load relationships with CSV Import.