Platform Control Center
Monitor fleet health, usage and operational events.
The control center is the single screen an operator needs to answer “is the platform healthy, who is growing, and what needs attention?”
Who this is for
The control center is available to platform staff. Regular workspace members see the tenant dashboard instead, which is scoped to their own data.
Control center sections
| Field | Type | Description |
|---|---|---|
| Headline metrics | string | Workspaces, accounts, data sources, pipelines, executions in the last 7 days and the all-time success rate. |
| Platform status | string | Database reachability and latency, engine, queued versus running executions, failures in the last 24 hours and tracked storage. |
| Execution volume | string | A 30 day stacked area chart of successful and failed runs across every workspace. |
| Outcome mix and busiest workspaces | string | All-time execution results as a donut, plus the workspaces generating the most runs. |
| Attention and adoption | string | Counters that need a follow-up, next to the plan mix across the fleet. |
| Execution activity | string | The most recent runs, newest first, with workspace, status, records and duration. |
Platform status
| Field | Type | Description |
|---|---|---|
| Database | string | Round-trip latency of a probe query, plus whether the connection succeeded. |
| Engine | string | Database backend in use, so you always know what you are tuning. |
| Queued runs | string | Executions waiting for a worker. Rising depth means you need more worker capacity. |
| Running now | string | Executions currently in flight. |
| Failures / 24h | string | Runs that ended in an error in the last day. |
| Uploaded sources | string | Raw files stored for data sources. |
| Materialised data | string | Processed datasets written by pipeline runs. |
| Tracked storage | string | Combined footprint of both, which is the number that grows. |
Capacity and usage
Use the status panel and the volume chart to answer two questions: is the fleet keeping up, and who is growing fastest.
- Queue depth rising while throughput is flat means you need more workers.
- Storage growth that outpaces record growth means datasets are getting wider, not longer.
- Busiest workspaces identifies who to contact before a limit is reached.
Attention and adoption
The attention grid lists the counters that usually need a follow-up. Every tile deep-links to the surface that resolves it, already filtered: suspended, planless and dormant workspaces open the workspace directory with the matching filter applied, locked accounts open the account directory, and the two run counters filter the execution feed below.
| Field | Type | Description |
|---|---|---|
| Suspended | string | Workspaces blocked from starting new runs. Open the workspace directory to restore them. |
| Without a plan | string | Workspaces with no assignment, which cannot run pipelines at all. Provision a plan to unblock them. |
| Dormant | string | Workspaces that have never completed a run — usually a stalled onboarding. |
| Locked accounts | string | Deactivated sign-ins. Reactivate from the account directory. |
| Queued runs | string | Executions waiting on a worker. A growing number is a capacity signal. |
| Failed in 24h | string | Runs that ended in an error in the last day. |
Next to it, the plan mix donut splits every workspace by the plan currently assigned, which is the fastest way to see how much of the fleet is on a paid tier.
Execution activity
The activity table is ordered newest first and shows the pipeline, workspace, status, record count, duration and start time for each run. Use it for the “did anything break in the last hour” question without leaving the control center.
Two controls narrow the feed: the status selector (queued, running, succeeded, failed or cancelled) and a time window of any time, the last 24 hours, 7 days or 30 days. Both are written into the URL, which is how the Queued runs and Failed in 24h tiles land on a pre-filtered feed. The heading reports how many runs match across the fleet, and the table shows the newest page of them.
Daily operating routine
- Check platform status for a degraded database or a rising queue.
- Review queue depth and failures in the last 24 hours.
- Clear the attention grid: plan, suspend and lock decisions first.
- Confirm no workspace has gone a full day without a successful run.
- Check tracked storage growth against the previous day.
Last reviewed October 2026.