Common Deployment Scenarios

This document presents the most typical scenarios in which Django Compressor can be configured, and should help you decide which method you may want to use for your stack.

In-Request Compression

This is the default method of compression. Where-in Django Compressor will go through the steps outlined in Behind the Scenes. You will find in-request compression beneficial if:

  • Using a single server setup, where the application and static files are on the same machine.

  • Prefer a simple configuration. By default, there is no configuration required.


  • If deploying to a multi-server setup and using COMPRESS_PRECOMPILERS, each binary is required to be installed on each application server.

  • Application servers may not have permissions to write to your static directories. For example, if deploying to a CDN (e.g. Amazon S3)

Offline Compression

This method decouples the compression outside of the request (see Behind the Scenes) and can prove beneficial in the speed, and in many scenarios, the maintainability of your deployment. You will find offline compression beneficial if:

  • Using a multi-server setup. A common scenario for this may be multiple application servers and a single static file server (CDN included). With offline compression, you typically run compress on a single utility server, meaning you only maintain COMPRESS_PRECOMPILERS binaries in one location.

  • You store compressed files on a CDN.

  • You want the best possible performance.


  • If your templates have complex logic in how template inheritance is done (e.g. {% extends context_variable %}), then this becomes a problem, as offline compression will not have the context, unless you set it in COMPRESS_OFFLINE_CONTEXT

  • Due to the way the manifest file is used, while deploying across a multi-server setup, your application may use old templates with a new manifest, possibly rendering your pages incoherent. The current suggested solution for this is to change the COMPRESS_OFFLINE_MANIFEST path for each new version of your code. This will ensure that the old code uses old compressed output, and the new one appropriately as well.

Every setup is unique, and your scenario may differ slightly. Choose what is the most sane to maintain for your situation.