Real-time population of dense volumetric data

Hi,
I’m developing an in-situ monitoring system for an additive manufacturing process. I currently do volumetric visualization of the builds after they are completed by exporting my data in RAW format and importing it into paraview for rendering. The data size can reach ~10GB so I’m using the client/server mode and running the rendering on my compute server that has sufficient memory and GPU power to do so.

I’m looking to set up a system such that my in-situ monitoring and detection system can dynamically update the data in real time on the server while the user is viewing it interactively via the Paraview client. I don’t want to save/reload the entire dataset from disk each time due to the size- only one z-axis layer is being modified at a time so the actual changes are quite small. I was hoping that there was some way to memory map the volume such that I can modify it without copying the full dataset.

The first step I’m looking at is about 1 update per minute, but I’d ultimately like to move to 60 updates per second if that is possible. I’ve been looking into Catalyst for this purpose but so far it seems geared for static rendering. Am I missing something here?

@nicolas.vuaille