Why 60,000 Config Items Should Not Be Cleared at Once
Once object counts move from dozens to tens of thousands, checking them one by one stops working: a single wrong deletion can affect production, and afterwards nobody can explain who changed what. Scale turns into a process problem — a ledger records what exists, an audit trail records who changed it.
Step 1: Inventory First, Then Build the Ledger
The inventory phase confirms counts, ownership and current state, then writes them into a ledger you can keep maintaining. For field design and how it maps to audit records, see this walkthrough of cleaning 60,000 config items.
- Object ID: a unique identifier to avoid name clashes
- Group: business line, environment, owner
- State: active, pending cleanup, archived
- Change basis: ticket number or approval record
- Last change time and operator
Device fleets put more weight on online status and batch ownership; trim the field list in managing devices at scale to fit your case.
Step 2: Execute in Batches, Log Every Batch
Do not aim to empty everything in one pass. Split the work by group or time window, then set a change limit and an observation period for each batch. The trail should capture time, operator, affected objects, values before and after, and the result — so a bad batch can be traced to a specific scope.
Step 3: Verify Results and Plan the Rollback
After each batch, spot-check the results before starting the next one, and keep a rollback path ready. For differences across many devices or environments, start with the comparison criteria in choosing a multi-device control platform.
Choosing a Tool: Ledger and Audit Trail First
- Batch push by group, with a cap per batch
- Automatic change records you can export
- Preview before execution and retry on failure
- Permission levels so no change is invisible
If you manage an account matrix instead of config items, the logic is similar; compare options in managing multiple social media accounts.