GlusterFS 1.2.1-BENKI Aggregated I/O vs NFSv4 Benchmark

From GlusterDocumentation

GlusterFS 1.2.1-BENKI Aggregated I/O Benchmark on 16 brick server and 64 clients over Infiniband SDP transport. Multiple 'dd' of varying blocks are read and written from multiple clients simultaneously.

NOTE: This benchmark is done for File I/O only. For metadata (stat(), readdir() etc) operations, benchmarks are not available.

Contents 1 GlusterFS Brick Config 2 NFS Server Config 3 Client Config 4 Interconnect Switch 5 GlusterFS 16 bricks (with ip-sdp transport) Vs NFS 6 GlusterFS performance with 64 clients and varying servers

GlusterFS Brick Config

• Processor - Dual Intel(R) Xeon(R) CPU 5160 @ 3.00GHz
• RAM - 8GB FB-DIMM
• Linux Kernel - 2.6.18-5+em64t+ofed111 (Debian)
• Disk: SATA-II 500GB
• Mellanox MHGS18-XT/S Infiniband HCA

NFS Server Config

• Processor - Dual Intel(R) Xeon(R) CPU 5160 @ 3.00GHz
• RAM - 8GB FB-DIMM
• Linux Kernel - 2.6.18-5+em64t+ofed111 (Debian)
• Disk: SATA-II 500GB
• Mellanox MHGS18-XT/S Infiniband HCA

Client Config

• Processor - Single Intel(R) Pentium(R) D CPU 3.40GHz
• RAM - 4GB DDR2 (533 MHz)
• Linux Kernel - 2.6.18-5+em64t+ofed111 (Debian)
• Mellanox MHGS18-XT/S Infiniband HCA

Interconnect Switch

• Voltaire 9288 96 port Infiniband Switch (14U)

GlusterFS 16 bricks (with ip-sdp transport) Vs NFS

Note that transfer rate is in BYTES per second (multiply by 8 for BITS per second).

Notes

• Single GlusterFS brick can perform better than single NFS server.

GlusterFS performance with 64 clients and varying servers

Notes

• With increased number of server, the overall performance increases linearly.