> because you cannot save all data (waste of time and resources) AFAIK, at least for ATLAS and CMS, the sheer amount of raw data the sensors produce, like 40TB

> because you cannot save all data (waste of time and resources) AFAIK, at least for ATLAS and CMS, the sheer amount of raw data the sensors produce, like 40TB/s or more[1], requires using triggers to reduce the captured data rate. Even after that first trigger there's still over 100GB/s, which is too much for storing directly. Interestingly though, due to its design the LHCb detector has a lower intrinsic luminosity (ie less stuff happening) and so was able to move to a fully software-based trigger system[2]. This allows them to change the trigger parameters "at will", unlike the hardware-based first-level triggers of ATLAS and CMS which can't really be changed much once installed. [1]: https://www.annualreviews.org/doi/full/10.1146/annurev-nucl-102115-044713 https://www.annualreviews.org/doi/full/10.1146/annurev-nucl-... [2]: https://arxiv.org/abs/1410.5012 https://arxiv.org/abs/1410.5012