I have been planning and testing running a video surveillance and
recognition system on a openssi cluster.
I have been able to solve the first small problems and got the system
running after a little tweaking. The application will be very bandwidth
intensive since it is collecting video streams from different cameras around
the country and then analyzing these streams for movement etc.
I am running openssi 1.2
The challenges I have encountered currently relate to shared memory
allocation. When I have only one node in the cluster everything works fine
but when I add nodes the applications starts to complain that the shared
memory is not enough. I have tried to modify it by increasing shmmax etc.
This got me thinking about the whole feasibility of the project. Now I have
been having problems when using 2-3 cameras. If I increase it to 100, will
the shared memory problems multiply or perhaps increase exponentially?
What do you more experienced users say?
Is this a totally crazy setup or do you think openssi could be used for
video capturing and analysing?
Thank you for your answers!