FAQ

What is Polars Nanook?

Polars is an open-source Python library for processing tabular data on a single machine. Polars Nanook runs the same Polars queries on remote machines in your own infrastructure, on one large machine or distributed across many. The overview shows an example.

Who is behind Polars Nanook?

Polars Nanook is built by the organization behind the open-source Polars library, which we continue to develop for single-machine workloads.

Where do my queries run?

In your own infrastructure, and your data stays there (see Security and data boundaries). Polars Nanook either starts EC2 instances in your AWS account when a query needs them, or runs on a Helm chart or bare-metal cluster you operate. Choose a deployment compares the deployments.

What does Polars Nanook add to the open-source Polars library?

The distributed engine, fault tolerance, compute on demand and query profiles, as described in What Polars Nanook adds.

When should I use Polars Nanook?

Use Polars Nanook when your data no longer fits in the memory of your machine, when a query needs more processing power than one machine has, or when your data is in cloud storage such as Amazon S3 and is too large to download.

The same query runs from a notebook, a scheduled job or an AWS Lambda function, so moving from experimenting to production needs no rewrite. If your data fits in the memory of a machine you already have, the open-source Polars library is enough.

Can Polars Nanook replace Spark for ETL?

Yes, for batch jobs that read, transform and write files in cloud storage. Migrate from Spark shows how and lists what does not move.

What does Polars Nanook cost?

The Developer plan is free. Team and Enterprise are paid plans. Plans and billing lists the prices and limits per plan and explains how vCPU-hours are counted.

Will the distributed engine be available in the open-source Polars library?

No, the open-source library remains single-machine. See How the distributed engine works.