---
title: "Performance Tips for Large Imports"
url: "https://docs.noreja.com/en/article/performance-tips-for-large-imports"
language: "en"
last_updated: "2026-10-23T08:28:57.694Z"
---

# Performance Tips for Large Imports

How an entity import works through its steps, and what makes it faster: fewer objects, properties and rows, indexes, simple joins and a quiet source system.

Large entities can take a while to import. This page explains how an import works through its steps and what you can do to make it faster. It's written for data specialists and their database administrators.

### How an Import Runs

-   An entity import has one step per object and one per data connection, followed by steps that build the model. The steps run one after the other.
-   Each object step reads its table from the source in pages of up to 250,000 rows, until all rows are read. Interfaces such as ServiceNow and OData are read in pages of 100 records.
-   A perspective import works on the data of the imported entity and doesn't read the source again.

During and after an import, the import details show each step with its start, duration and the number of imported items; see [Import Entities and Perspectives: Status and Errors](https://docs.noreja.com/en/article/import-entities-and-perspectives-status-and-errors). Use them to find the steps that take longest.

### Read Less

-   **Only the objects you need**: every object and every data connection is a step, whether a perspective uses it or not.
-   **Only the properties you need**: every selected column is read for every row.
-   **Only the rows you need**: limit objects with a WHERE clause, for example to the last two years, or to the document types the process covers. See [Configure Objects and Relationships](https://docs.noreja.com/en/article/configuring-an-entity).
-   **Several smaller entities**: if a data source holds several processes, give each its own entity. See [Entity Modelling Best Practices](https://docs.noreja.com/en/article/entity-modelling-best-practices).

### Make Reading Faster

-   Index the columns used in data connections, JOIN clauses and WHERE clauses, and the key columns. See [Prepare Your Source Data: Keys, Timestamps, Indices](https://docs.noreja.com/en/article/source-data-preparation-common-challenges), including the Smart Index Suggestion tool.
-   Prefer simple JOIN clauses. Every join is part of the query of each page. See [Use Joined Properties](https://docs.noreja.com/en/article/use-joined-properties).
-   Make sure the source database isn't busy with other heavy jobs during the import, for example backups or nightly batch runs.
-   For Snowflake, choose a warehouse size that fits the amount of data. See the section Prepare Snowflake in [Connect a Data Source](https://docs.noreja.com/en/article/connecting-data-sources).

### When an Import Fails

If a step fails, the import details show the failed step and the error message; **Copy error message** copies it for your administrator or Noreja's support. Long-running queries in the Builder, such as a test query, stop after three minutes; requests to interfaces after 50 seconds, with a few retries. See [Import Entities and Perspectives: Status and Errors](https://docs.noreja.com/en/article/import-entities-and-perspectives-status-and-errors) and [Troubleshooting Imports](https://docs.noreja.com/en/article/troubleshooting-imports).

### Related

-   [Entity Modelling Best Practices](https://docs.noreja.com/en/article/entity-modelling-best-practices)
-   [Prepare Your Source Data: Keys, Timestamps, Indices](https://docs.noreja.com/en/article/source-data-preparation-common-challenges)
-   [Import Entities and Perspectives: Status and Errors](https://docs.noreja.com/en/article/import-entities-and-perspectives-status-and-errors)