We are planning to go for data storing in ROOT format for an experiment. But we are struggling with an issue.
Question 1) whether to save the acquired data directly in ROOT format in Data acquisition node Or temporarily defer, saving the data in binary format, then converting the binary data to ROOT on a different thread. The problem we face if we directly save into ROOT format is: network throughput is being limited by the time it takes to save into ROOT TTree. So we are planning to go for binary file writing, which we observed it faster than ROOT TTree writing. Can you comment, which is the right approach for data saving on data acquisition node?
Question 2) Normally if we consider High Energy Experiements, how do they save the acquired initially? Do they directly save the acquired data directly into ROOT format or else they dump the binary data later convert it ROOT format?
Some more details:
The network throughput we expect in worst case is: 1Gbps
Event Data ~= 9000 bytes
Any suggestion in this regard is really appreciated.