---
title: "Data collection"
description: "When Curade collects each kind of data from a tenant's sources, how it fills gaps, how long it keeps history and what pauses collection."
url: https://curade.ai/en/curade/docs/concepts/data-collection/
product: "Curade"
version: "3.0.0"
breadcrumbs:
  - "Curade docs"
  - "Concepts"
  - "Data collection"
---

# Data collection

When Curade collects each kind of data from a tenant's sources, how it fills gaps, how long it keeps history and what pauses collection.

Curade collects each tenant's data on a schedule, from every active source of every active tenant. All the times below are Malaysia time.

## What is collected, and when

| Data | When |
|---|---|
| Inventory of hosts, applications and services | Once a day, at 00:05 |
| Host and disk readings, from most sources | Every minute, with a second pass every 15 minutes that fills any gap in the last half hour |
| Host and disk readings, from sources read hourly | At five past each hour, with a second pass at twenty past that fills any gap |
| Transactions | Once a day, at 02:00, for the day before |

A transaction you register is filled in with the history your organisation's monitoring platform still holds, so it does not start empty.

## When a source fails

Each source is collected on its own, so one that fails, with an expired token for example, never stops the others. The next pass collects what the failed one missed, within the stretch it covers.

## How long history is kept

Host, disk and transaction history is kept for seven years, so dashboards, capacity planning and reports can look back over months and years.

## What pauses collection

Collection needs both the tenant and the source to be active. Switching either to inactive stops collection and keeps the history; switching it back resumes it from then on ([Tenants and sources](https://curade.ai/en/curade/docs/concepts/tenants-and-sources/)).