Paraview is a visualization package for large data sets. It overcomes the issue of poor performance of graphics over a network by splitting the processing and rendering over multiple servers. In this model, paraview can connect to your data on a remote system, process the graphics elements on the remote system, but render and display the graphics on your local workstation, taking advantage of hardware acceleration. In addition to having the processing done on a single server, paraview can use MPI to connect to multiple server nodes to process very large data sets. Finally, paraview can submit jobs to a queue system on the processing servers. Typically, the only way to access compute nodes on an HPC system is through a queue system so this allows paraview to take advantage of those resources.
Paraview can be a little difficult to set up, particularly on Windows machines. First off, you must have passwordless SSH setup. There is no way to type a password when Paraview connects so this step is crucial. Second, you must pass Duo 2-factor credentials. Since there is no prompting the DUO_PASSCODE variable must be set in order to pass a code to Duo when Paraview connects via ssh. This can be handled via the paraview connection settings for Mac and Linux, but not on Windows. Windows systems require a hardcoded value be set prior to launching Paraview. This can be done in Putty by setting the DUO_PASSCODE variable in the session section of the settings. See the Two-factor authentication page for details.