Rollups and Analytics

Chirpier exposes two different ways to read an event.

Rollups

Use event logs when you need the raw time-bucketed history for an event.

Supported rollup periods:

minute
hour
day

Endpoint:


GET /v1.0/events/:eventID/logs?period=minute|hour|day&limit=&offset=

What rollups store

Each rollup bucket aggregates all logs that fell within that time window:

Field	Description
`count`	Number of logs in the bucket
`value`	Sum of all `value` fields
`min`	Smallest value logged
`max`	Largest value logged
`squares`	Sum of squared values (used to compute variance and standard deviation)

Chirpier stores these pre-aggregated buckets rather than raw log data. This is how it supports average, p95_est, and p99_est aggregates in policies without retaining every individual log long-term.

Rollups are best for:

charts
recent history inspection
building your own comparisons
validating policy inputs

Analytics windows

Use analytics windows when you need Chirpier to compare the current window to a previous one.

The response schema stays the same for every analytics window. current_* always refers to the requested window, and previous_* always refers to the comparison window selected by the previous query parameter.

Endpoint:


GET /v1.0/events/:eventID/analytics?view=window&period=...&previous=...

Supported analytics periods:

1h
1d
7d
1m

Supported comparison keys:

previous_window
previous_1d
previous_7d
previous_1m

Common comparison examples:

period=1h&previous=previous_window — this hour vs the previous hour
period=1h&previous=previous_1d — this hour vs the same hour yesterday
period=1d&previous=previous_7d — today vs the same day last week
period=7d&previous=previous_window — this week vs last week
period=1m&previous=previous_window — this month vs last month

For the full combination table, see the Analytics API reference.

Analytics windows return a comparison summary rather than a time series. Typical response fields include:

current_value, current_count, current_mean, current_stddev
previous_value, previous_count, previous_mean, previous_stddev
value_delta, count_delta, mean_delta
value_pct_change, count_pct_change, mean_pct_change

Choosing between them

Use rollups when you need the time series.

Use analytics windows when you need a precomputed comparison summary.

For route-level details, see Event logs and Analytics API.