We are looking to develop a Client/Server configuration using ParaView server and one of the available ParaView web technologies. The datasets we have are medium sized and we anticipate multiple simultaneous sessions. These sessions will be initiated by various divisions (Tenant) in our global company. We want to ensure that each divisions work is running completely secure and isolated from others. Currently, we plan to run a single cloud-based VM initially, with the intention to scale as demand increases. Hence, I am looking for advice on the following:
What would be the best Web Technology to Use?
What is the best way to configure the server so that competing workloads generated by each division (tenant) are robust (i.e. don’t cause ParaView to crash) and are less vulnerable to noisy-neighbor problems? For instance, do I need to use some form of virtualization, such as containers, or will ParaView handle that?
Trame, but to be fair, I would say it depends. Sometime it is better to have a couple of back and forth to exactly grasp what are your needs, where you are coming from (company constraints/previous investment/…) and what you plan to do in the future.
I would say that could depend on your setup and how you want to manage the load. Tools like nomad could provide some help if you already know the expected load of each job before they start. Basically a virtualization infrastructure will help.
HTH,
Seb
PS: those are hard questions to answer and it might be good to have a deeper conversation.
Sorry for the late reply. If I were to use client side rendering where the geometry is sent to the client is the browser able to use GPU? Also this blog
So when we do local rendering (geometry exchange), then yes, the client use its GPU to do the rendering.
I don’t understand states Do not use local rendering with actual datasets.
But you can do local rendering for production. It always boils down to geometry size, network latency and client capability. For some it will work great, for other not so much…