# How data arrives

Understand where the numbers you read come from, so you know which route to choose and what to expect from each.

## Four ways to feed Naoshift

| Route | What it is | What it is for |
|---|---|---|
| **Collector** | A small service on a machine in your network, polling your hypervisors at regular intervals | Continuous tracking: history, changes, capacity over time |
| **File import** | An RVTools or LiveOptics export dropped on the Integration page | A first look in minutes, nothing to install |
| **Cloud connector** | A connection to Microsoft Azure, with no agent | Your cloud machines in the same inventory |
| **Backup connector** | Veeam, Commvault, Proxmox Backup Server, Rubrik, read through their API | Protection matched to each machine |

## Source and collection

Each route is a **source**: a vCenter, an Azure subscription, a lineage of RVTools files, a backup server. A company usually has several.

Each pass of a source produces a **collection**, a snapshot of what it sees at that moment. Naoshift compares two collections of the same source to tell what changed.

## The consolidated current state

Current-state screens show the merge of the **latest healthy collection of each active source**. That is the consolidated current state: your estate as it is now, across all sources.

> Two sources that see the same machine count it twice. So declare one entry point per scope: the vCenter, rather than the vCenter and each of its hosts.

## Anonymization

If it is enabled for your company, the names of machines, hosts and clusters and IP addresses are replaced by tokens **during the import**. Analyses and the assistant work on those tokens. You reveal the real names on screen with the eye icon.

## Which route to choose

- You want an immediate overview, without opening flows: [import a file](/en/connect/import-file).
- You want to follow your estate over time: [prepare a collector installation](/en/connect/prepare).