For YCSB, we evaluate with Workloads A, B, and C. For detailed configuration and setup, please refer to our userguide.

YCSB

Workload A

swl

Workload B

srl

Workload C

srl

Experimental Testbed: Each compute node in this cluster has two twelve-core Intel Xeon E5-2680v3 processors, 128GB DDR4 DRAM, and 320GB of local SSD with CentOS operating system. Each node has 64GB of RAM disk capacity. The network topology in this cluster is 56Gbps FDR InfiniBand with rack-level full bisection bandwidth and 4:1 oversubscription cross-rack bandwidth.

These experiments are run with 4 Regionservers and upto 16 client nodes, each running 1-8 HBase client instances. Each client uses 1 thread for communication and performs 25000 operations on the HBase cluster. Before running the workload, we load 25000 rows in an HBase table and run the workload on that table. The RDMA-IB design improves the throughput for Workload A, B, and C by up to 145.51%, 164.82%, and 259.41% over IPoIB (56Gbps), respectively.