Distributed Workflows and Remote Rendering

A Vistle workflow can be distributed between connected clusters. This can be useful for several reasons:

• The data resides on a different system than where the visualization shall be displayed

• A certain module is only available on one system (e.g. because of licenses or architectural differences)

In such a scenario, modules running on different cluster can be connected just as when running on a single cluster. Additionally, remote rendering integrates seamlessly with distributed workflows.

Setting Up a Distributed Session

The first instance of Vistle that is started includes the primary hub. All other instances/hubs have to connect to this instance. If establishing TCP connections is hindered by e.g. packet filters/firewalls or network address translation (NAT), then you should start the primary instance on a host that every other participating system can connect to. The primary system will show on which port it can be reached. The default is 31093. Secondary instances can be spawned with this command:

vistle -c primary.host.name:31093