RendD is the actual rendering service for the distributed rendering system. It can be started directly or through RendW program.
RendD program must be running on every workstation which is acting as a rendering server. The program listens incoming rendering requests and starts rendering threads to take advantage of the processor power available in the workstation.
The program can be used for distributing rendering over any TCP/IP based network, such as the Internet.
In addition to local threads, RendD can distribute rendering to other rendering daemons running on other workstations. This means that hierarchical rendering farms can be configured.
In distributed network rendering, all data of the rendered project needs to be sent to all rendering servers. Transmitting data becomes often the bottleneck especially when one needs to distribute rendering over slow networks (which may be the case when distributing over the Internet). By using hierarchical rendering networks, the client sends projects only to a small number of servers, which then distribute the task further to specified sub daemons of a fast LAN etc.
Two methods exist for setting up hierarchical rendering farms. Sub rendering daemons can be specified 'manually' as startup parameters. Another alternative is to the Automatic Sub Servers option, which automatically distributes rendering to the workstations available in LAN or registered to it. Automatic approach has some advantages, because it makes the configuration of the rendering farm flexible - it can change any time.