Overview
This dashboard shows incident time-to-resolution (TTR) for every monitored segment (database.table, optionally a client feed id), grouped by table. It is a companion to the Data Trust monitor: the same Monte Carlo freshness data, but focused on how long late deliveries take to recover rather than current status.
An incident is one recovered freshness breach: the table crossed its freshness SLA (went stale) and later refreshed. Its TTR is the downtime beyond the SLA — from the moment the deadline was breached to the moment the data recovered. This is the same quantity the Data Trust dashboard averages as MTTR, shown here per incident.
Freshness status
Every segment (and its table / domain / feed roll-up) carries its current Monte Carlo freshness status. Status is a point-in-time state and does not change with the TTR window. The summary bar totals it across everything shown.
Fresh
The latest observation is within the freshness threshold — the table updated on time.
At Risk
Still Fresh, but a source pipeline failed or missed its start SLA today, the data has not refreshed today, and it is due to go stale within the next 12 hours. Hover the badge for the pipeline detail.
Stale
The latest observation breached the freshness threshold — hours since last update exceeds the allowed threshold.
Training
Monte Carlo has not learned a freshness threshold for this segment yet, so Fresh / Stale and TTR don't apply. These have no incidents to plot, so they are counted in the summary but not charted here.
Time window
The 7 / 30 / 90 day toggle re-scopes every chart to incidents that resolved within that window. All statistics (n, median, mean, gross, p90, worst) are recomputed for the selected window.
Uptime %
Each segment shows its Uptime % for the selected window — healthy time ÷ total time across its freshness observations (a table is "healthy" until it crosses its SLA deadline). Colored by band: ≥ 99.5% good, ≥ 95% at risk, below that poor. Each group rolls up to the median of its segments' uptime. You can also sort by uptime (lowest first).
Group by
Re-group the segments without changing the TTR view or window:
Domain
Segments grouped by database — each domain is a header band over its table cards (default).
Table
A flat list of table cards — the same grouping as the Data Trust Flat view.
Feed ID
Segments grouped by client feed id; each card shares one x-scale across that feed's tables.
Views
Median TTR
Horizontal bar per segment — the typical recovered incident length (p50). Robust to a single very long outage. A currently-ongoing breach has no final TTR and is excluded (it still counts toward Gross and Worst).
Mean TTR
Horizontal bar per segment — the average recovered incident length. Pulled upward by long outliers. Excludes the current ongoing breach, matching the Data Trust dashboard's MTTR.
Gross TTR
Horizontal bar per segment — total downtime (sum of every incident's TTR) in the window. The overall reliability cost.
Detail
One horizontal dot strip per segment — every incident is a dot placed by its TTR, with a small vertical jitter so overlapping incidents stay visible. Summary lines are drawn by sample size: n < 5 → dots + worst marker only; 5 ≤ n < 10 → dots + median line + worst; n ≥ 10 → dots + median line + p90 line.
Low-sample segments (n < 5)
Segments with fewer than 5 incidents in the window carry too little data to trust a summary statistic. In the bar views they render as a dotted, semi-transparent bar. The Show n < 5 toggle hides or reveals them across every view.
Controls
Search
Real-time filter across segment keys, table names, clients, business lines, feed names, and feed types.
Clients / Business Line / Feed Type
Multi-select dropdowns to scope to specific clients, business lines, or feed types (feed-level segments).
Sort
Order tables by Worst First (most total downtime), Most Incidents, Table Name, or Segment Count. Click an active sort to reverse.
Row breakdown
Left-click any segment row to expand a breakdown of its individual incidents in the current window. Sort the breakdown by Latest (most recent first) or Max downtime (longest TTR first).
Loading data
The latest snapshot is fetched automatically on load. You can also drag any .json Incident Detail snapshot onto the page to load it immediately.