ODOCK.AI
ObservabilityTraffic Analytics

Filters and scope

The filter bar and the sampling controls that shape every chart on the page.

Filters and scope

The filter bar at the top of the Traffic Analytics page constrains every KPI card and chart on the page. Sampling controls how much data the page loads.

Filter bar on the Traffic Analytics page

Filter Controls

ControlUse it to
RangeRestrict to the last 24h, 7d, 30d, 90d, or everything loaded.
GatewaySwitch between AI and MCP, AI only, or MCP only.
EntityPick the dimension used for entity breakdowns: provider, model, team, user, API key, MCP tool, or MCP server.
Entity valueRestrict to one specific value of the selected entity.
StatusAll, success, errors, or unknown only.
SearchFree-text match against request id, trace id, provider, model, principal, error code, MCP tool, or MCP server.

Filter changes apply instantly — no extra load.

Sampling

ControlUse it to
Max recordsCap the number of records pulled. Useful on long ranges.
SampleThe percentage of available records sampled by the backend. Lower percentages mean a faster but coarser page.

The KPI cards and the footer text always reflect the sampled slice, not the full population. When you need exact counts, use the Usage Records analytics card instead.

Refresh

The Refresh button on the filter card forces a reload with the current settings. The page shows an inline error message if the reload fails and keeps the last successful view on screen.

Tips

Set Entity to API Key and pick a single key to baseline one application. The latency and error KPIs then reflect that key only — that is the fastest way to compare a key against itself after a deploy.

Sampling is non-deterministic — two viewers with the same filters may see slightly different rows. Use Usage Records when reproducibility matters.

On this page