Workload fleet

Route: /workloads

Purpose

The fleet cockpit provides a portfolio view of active workloads. Use it to identify unknown or low-health workloads, see estate composition, compare environment and criticality, and open a workload for deeper analysis. Workload fleet cockpit showing health and resource composition

When to use it

Daily or weekly fleet triage.
Finding workloads that have never been analyzed.
Prioritizing production/critical workloads with coverage or retirement risks.
Starting Autopilot, manual workload creation, merge, grouping, or overlap analysis.

Prerequisites and data sources

Prerequisites and permissions

workloads.read to view the fleet and profiles.
workloads.write to create, edit, merge, delete, or group workloads.
Existing workloads, or a readable Azure connection for discovery.
Prior feature scans for meaningful health and risk values.

Tabs and actions

Surface	Actions
Fleet	Switch cards/table/board/map, group cards by application, filter by group/classification/search/resource count, sort, save a layout/filter view, select all shown
Selection toolbar	Launch durable missions, launch deep reviews, merge, group, or move up to 500 workload definitions to Trash
Workload card/row	Open, refresh, chat, deep review, mission, assess, edit, or move one definition to Trash
Trash	Restore one workload, permanently purge one, or permanently empty Trash
Discovery and integrity	Open Autopilot, Groups, Overlaps, or estate-coverage orphan triage

The Resources minimum and maximum are inclusive. The count uses the cache-only workload profile’s composition total when available; before a profile is available it falls back to the number of explicit resource nodes in the workload definition. The filter applies to cards, table, board, map, grouped cards, and Select all shown, and its values persist in the browser.

Freshness and scope behavior

The active workload list and Trash are application-registry reads. Fleet profiles, group rollups, overlap summaries, health and risk are cache-only; opening the page does not scan Azure. Profiles and lightweight group/overlap queries are treated as fresh for 60 seconds in the browser.

The workload definitions can arrive before the shared fleet-profile response. During that initial gap, each card keeps its name and actions visible while a sliding shimmer and fixed-size skeletons occupy the score, composition and health-visualization areas. This is a loading state, not a claim that the workload has zero resources. The shimmer stops when profiles succeed or fail, does not cover already-cached cards during a background refetch, and becomes static when the browser’s reduced-motion preference is enabled.

Selection is keyed by workload ID rather than by the current view. Switching layouts or narrowing a filter does not discard earlier selections, so the selection count can exceed the visible row count. Opening Trash clears the active selection to prevent lifecycle actions crossing views.

Workflow overview

Workflow

Open /workloads.
Review fleet composition, environment-by-criticality distribution, health bands, and risk indicators.
Filter or sort to place critical, production, unknown, or poor-health workloads first. The Resources min–max filter isolates fragmented definitions (for example, max 4).
Open a workload card or row to inspect its detail.
For Not analyzed, run analysis from workload detail rather than assuming failure.
Use Autopilot to discover missing application boundaries.
Open Overlaps when resource ownership is ambiguous, or Groups to compare related environments.
Launch Mission Control for a coordinated multi-system sweep.

Bulk lifecycle operations

The selection toolbar works across cards, table, board, map, grouping, search, and filters. Use Select all shown to add the current visible result while retaining selections made elsewhere. Move N to Trash soft-deletes up to 500 selected workload definitions in one operation. The confirmation names the workload/resource totals and any group memberships.

Bulk Trash does not modify Azure resources and does not cascade-delete architectures, Know-Me, FMEA, assessments, missions, evidence, or audit history. It reports workloads moved, already trashed, or no longer found. Restore from Trash before permanent purge if the selection was wrong.

Safety and lifecycle

Editing membership changes the scope used by downstream analyses.
Merge moves source workloads to Trash and has no dedicated undo; review members and downstream links first.
Normal delete is soft-delete and can be restored. Purge and empty-trash are permanent.
Trashed workloads are excluded by downstream active-workload consumers. Their group association metadata is preserved so restoration returns them to the same group.
Fleet pages read cached profiles and should not be mistaken for a live Azure scan.

Interpretation of results

Interpret fleet health

A workload profile combines available monitoring, telemetry, backup/DR, performance, ownership, policy, and tag signals. The overall score reweights only the signals that are present.

Band	Score	Meaning
Good	80–100	Available signals are broadly healthy; inspect freshness and any remaining risks
Warning	50–79	Material gaps exist and should be planned for remediation
Poor	0–49	Available signals indicate significant gaps or failures
Not analyzed	No score	No usable signals have been computed; this is unknown, not zero

Because the score is normalized over present signals, two workloads with different signal coverage are not always directly comparable. Open detail and check freshness and component scores.

Exports, history, scheduling, and integrations

No dedicated export, history, scheduling, or integration controls are documented for this feature page.

Safety and limitations

Resource-count filtering is a browser view filter, not a deletion policy. Review the selected count and confirmation names before bulk Trash, especially after filters have hidden rows.
Bulk Trash accepts 1–500 distinct workload IDs and performs one local registry write. Missing or concurrently trashed IDs are reported rather than rolling back definitions that were moved.
Trash changes only workload definitions. Azure resources are never deleted or moved.
Group association metadata is preserved in Trash and returns on restore. Purge and empty Trash permanently remove definitions and have no restore path.
Resource counts can lag Azure until a workload is refreshed or edited; inspect membership before treating the count as authoritative.

Troubleshooting

Symptom	Resolution
Fleet is empty	Run Autopilot or create a workload; verify `workloads.read`
Workload says Not analyzed	Open it and run Analyze
Score changed sharply	Compare component freshness and determine which newly available signal changed normalization
Resource count is stale	Refresh an Autopilot-origin workload or edit its nodes
A deleted workload vanished from the active group view	Restore it; active views exclude Trash, but the saved group association returns with it
Bulk Trash action is disabled	Select 1–500 active workloads; more than 500 must be split into multiple reviewed batches
Merge result is unexpected	Inspect the merged workload and source entries in Trash before any permanent purge
Resource filters show no workloads	Clear the minimum/maximum values or correct a minimum greater than the maximum. Both boundaries are inclusive.
Cards briefly show a sliding highlight	Fleet profiles are still loading from the server cache. Wait for the visualization skeletons to resolve; no Azure scan is running.
A card says Visualization unavailable	The shared cached-profile request failed. Reload to retry; the workload definition and its actions remain available.
Selected count is larger than the visible result	Selection persists when layouts or filters change. Use Clear or Deselect all and review the selected count before Trash.
Bulk Trash moved fewer definitions than selected	Some IDs were already trashed or disappeared concurrently. Read the result banner and reopen Trash before retrying.