Client - Repository - Webserver

As you can see in the architecture overview there is basically a 3-tier concept behind IWS:

Client

The client connnects to the repository server to edit documents, to add languages, to trigger a deployment or what ever. Ths client is your interface to IWS.

Repository

The repository holds all files and documents and executes most of the workload. The deployment is always executed on the repository server - not on the client. The repository server just gets triggered by the user (via the IWS client) and starts the deployment.

Webserver

The repository server connects to the webserver and uploads all files and documents via FTP (or does a local copy, if the webserver is on the same machine as the repository server). The repository server keeps a list of files that are already on the webserver, so that onyl changed files get deployed (unless "All files" is selected).

IWS can handle as many webservers as you want. A common example is to configure one production webserver location and one staging location, to test major changes before they get published to the public. This staging location may still reside on the same physical webserver as the production server, you may choose a different subdirectory or better an other virtuual directory.