Sherman | Optimized Distributed B+Tree Index

You can modify this line according the RDMA NIC you want to use, where ibv_get_device_name(deviceList[i]) is the name of RNIC (e.g., mlx5_0) https://github.com/thustorage/Sherman/blob/9bb950887cd066ebf4f906edbb43bae8e728548d/src/rdma/Resource.cpp#L28. If you use RoCE, modify gidIndex in this line according to the shell command show_gids, which is usually 3. https://github.com/thustorage/Sherman/blob/c5ee9d85e090006df39c0afe025c8f54756a7aea/include/Rdma.h#L60. If you use RoCE and the MTU of your NIC is not equal to 4200 (check with ifconfig), modify the value path_mtu in src/rdma/StateTrans.cpp. Change the constant kLockChipMemSize in include/Commmon.h, making it <= max size of on-chip memory.
For each run with kNodeCount servers:. We emulate each server as one compute node and one memory node: In each server, as the compute node, we launch kThreadCount client threads; as the memory node, we launch one memory thread. kReadRatio is the ratio of get operations. In ./test/benchmark.cpp, we can modify kKeySpace and zipfan, to generate different workloads. In addition, we can open the macro USE_CORO to bind kCoroCnt coroutine on each client thread.
cd Sherman
./script/hugepage.sh to request huge pages from OS (use ./script/clear_hugepage.sh to return huge pages)
mkdir build; cd build; cmake ..; make -j
cp ../script/restartMemc.sh .
configure ../memcached.conf, where the 1st line is memcached IP, the 2nd is memcached port
./restartMemc.sh (to initialize memcached server)
In each server, execute ./benchmark kNodeCount kReadRatio kThreadCount