Work Project Report: Benchmark for 100 Gbps Ethernet network analysis

During the Long Shutdown 2 (2018-2019) the LHCb experiment will be upgraded in order to reach extremely high precision on the main observable of the b and c-quarks sectors. In its current state the LHCb experiment has a readout rate of 1.1 MHz within a fixed latency, which causes the collision rate to be limited. The removal of this bottleneck is one of the main objectives of the LHCb upgrade. This will be achieved by implementing a trigger-less readout system where the functionality of the trigger will be executed by software. The trigger-less readout system must be able to handle large bandwidth such as 4 TBytes/s. For that reason the design of the system will include readout boards that can transmit information at the rate of 100 Gbits/s and a high throughput local area network [1]. The main goal of my project is to create an automatized benchmark for automatic launching, scheduling and analysing the performance of the readout boards and the network, in order to decide which manufacturer offers the best solution for our implementation, what configuration of the network offers the best performance and what configuration of the readout system is the most cost efficient.