> ## Documentation Index
> Fetch the complete documentation index at: https://documentation.kodelabs.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Table structures and polling

> Understand how Building BI time-series sources record rows, including COV, structured pulls, and aggregated polling.

Choose a time-series source in [Data sources](/products/building-bi/data-sources#key-data-sources). Polling behavior is usually implied by the source name (Real-Time, Last, Hourly, Daily).

Every point time-series table holds **one row per point per time period**, plus building, area, and device context. The same device can produce multiple rows at one timestamp — one per point.

## Primary structures

| Structure                    | Type                  | Behavior                                             |
| ---------------------------- | --------------------- | ---------------------------------------------------- |
| Point Time Series Real-Time  | Physical (ClickHouse) | Every reading is a new row. Full history.            |
| Point Time Series Last Value | Virtual               | Most recent value per sensor. Timestamps may differ. |

Use Last Value to compare current state across sensors with misaligned poll rates. See [Real-Time vs Last Value](/products/building-bi/data-sources#key-data-sources).

## Polling types

| Polling type                    | Behavior                                                                                      |
| ------------------------------- | --------------------------------------------------------------------------------------------- |
| Change of Value (COV)           | New row when the value changes past the COV threshold, with a **1-hour maximum** between rows |
| Structured Data Pull            | One row per interval (for example every 1, 5, or 15 minutes)                                  |
| Aggregated (for example Hourly) | Sum, Min, Max, Avg, or Count per point per period                                             |

Aggregation suits some points and not others. Hourly average air temperature is useful; aggregated elevator floor level is not.

Prefer reduced-frequency sources over Real-Time only when load time is a concern.

## Next steps

<CardGroup cols={2}>
  <Card title="Key data sources" icon="database" href="/products/building-bi/data-sources#key-data-sources" arrow={true}>
    Pick Real-Time, Last Value, or reduced-frequency sources.
  </Card>

  <Card title="Column reference" icon="columns-3" href="/products/building-bi/data-sources#column-reference" arrow={true}>
    Look up Point Time Series column names.
  </Card>
</CardGroup>