SPECsfs2008_nfs.v3 Result

Hitachi Data Systems : Hitachi Unified Storage File, Model 4100, Four Node Cluster
SPECsfs2008_nfs.v3 = 607647 Ops/Sec (Overall Response Time = 0.89 msec)


Performance

Throughput
(ops/sec)
Response
(msec)
60571 0.2
121278 0.3
182106 0.4
243041 0.5
303870 0.6
364670 0.7
425747 0.9
487902 1.1
548184 1.5
607647 5.5
Performance Graph


Product and Test Information

Tested By Hitachi Data Systems
Product Name Hitachi Unified Storage File, Model 4100, Four Node Cluster
Hardware Available July 2013
Software Available September 2013
Date Tested August 2013
SFS License Number 276
Licensee Locations Santa Clara, CA, USA

The Hitachi Unified Storage (HUS) and Hitachi NAS (HNAS) Platform family of products provide multiprotocol support to store and share block, file and object data types. The 4000 series delivers best-in-class performance, scalability, clustering with automated failover, 99.999% availability, non-disruptive upgrades, smart primary deduplication, intelligent file tiering, automated migration, 256TB file system pools, a single namespace up to the maximum usable capacity, and are integrated with the Hitachi Command suite of management and data protection software. The HUS file module uses a hardware-accelerated "Hybrid Core" architecture that accelerates network and file protocol processing to achieve the industry's best performance in terms of both throughput and Operations per second. The HUS file uses an object-based file system (Silicon File System) and virtualization to deliver the highest scalability in the market, enabling organizations to consolidate file servers and other NAS devices into fewer nodes and storage arrays for simplified management, improved space efficiency and lower energy consumption. Each 4100 node or cluster can scale up 16PB of usable data storage and support 10GbE LAN access and 8Gbps FC storage connectivity.

Hitachi Unified Storage VM includes flash optimized system software and patented Hitachi Accelerated Flash to improve application performance. Utilizing external storage virtualization and automated tiering, HUS VM centralizes storage management of multiple storage tiers for the highest economic value.

Configuration Bill of Materials

Item No Qty Type Vendor Model/Name Description
1 4 Server HDS SX345384.P Hitachi NAS 4100 Base System
2 1 Server HDS SX345278.P System Management Unit (SMU)
3 1 Software HDS SX365131.P Hitachi NAS SW - Software Bundle
4 16 FC Interface HDS FTLF8528P3BNV.P SFP+ 8Gbps FC
5 24 Interface HDS FTLX8571D3BCV.P SFP+ 10GE
6 1 Storage HDS HUS-VM-SOLUTION.S HUS VM Storage Platform
7 1 Storage HDS HUS-VM-A0001.S HUS VM Product
8 1 Chassis HDS DW700-CBX.P HUS VM Controller Chassis
9 16 Cache HDS HDW-F700-16GB.P HUS VM 16GB Cache Module
10 1 Cache HDS DW-F700-BM256.P HUS VM Cache Flash Memory Module (in use only during power outage)
11 64 Drives HDS HDW-F700-1R6FM.P HUS VM 1.6TB Flash Module Drive
12 6 Drive Chassis HDS DW-F700-DBF.P HUS VM Drive Box (Flash)
13 4 FC Interface HDS HDW-F700-HF8G.P HUS VM 4x8Gbps FC Interface Adapter
14 4 Disk Adapter HDS DW-F700-BS6G.P HUS VM B/E I/O Module
15 1 Rack HDS A3BF-SOLUTION.P Solution 19 in rack ROW MIN
16 5 Software HDS 304-232001-03.P SVC Mo HUS VM Hitachi BOS Base Lic (20TB) - SW Sppt
17 2 Switch Brocade HD-5320-0008.P Brocade 5320 FC Switch
18 2 Switch Brocade TI-24X-AC.P Brocade TurboIron 24x 10GbE switch for Cluster Interconnect

Server Software

OS Name and Version 11.2.3319.09
Other Software None
Filesystem Software SiliconFS 11.2.3319.09

Server Tuning

Name Value Description
security-mode UNIX Security mode is native UNIX
cifs_auth off Disable CIFS security authorization
cache-bias small-files Set metadata cache bias to small files
fs-accessed-time off Accessed time management was turned off
shortname off Disable short name generation for CIFS clients
read-ahead 0 Disable file read-ahead

Server Tuning Notes

None

Disks and Filesystems

Description Number of Disks Usable Size
1.6TB Hitachi Accelerated Flash module drive 64 89.6 TB
250GB SATA Disks. These eight drives (two mirrored drives per node) are used for storing the core operating system and management logs. No cache or data storage. 8 1000.0 GB
Total 72 90.5 TB
Number of Filesystems 8
Total Exported Capacity 89.6TB
Filesystem Type WFS-2
Filesystem Creation Options 4KB filesystem block size
Filesystem Config Each file system was striped across 8 LUNs from a single 7D+1P, RAID-5 Group consisting of 8 Flash module drives.
Fileset Size 70907.5 GB

The storage configuration consisted of one Hitachi Unified Storage VM All Flash storage system (HUS VM) configured with single chassis and 256GB cache memory. There were 64 1.6TB Hitachi Accelerated Flash module drives in use to meet the capacity and performance requirements of the benchmark. There were 64 LUNs created using RAID-5, 7D+1P. There were sixteen 8Gbps FC ports in use across two FED features located in different clusters. The FC ports were connected to the 4100 nodes via a redundant pair of Brocade 5320 switches. The 4100 nodes were connected to each Brocade 5320 switch via two 8Gbps FC connections, such that a completely redundant path exists from each node to the storage. Each Hitachi Unified Storage file module nodes have two internal mirrored hard disk drives that are used to store the core operating software and system logs. These drives are not used for cache space or for storing data.

Network Configuration

Item No Network Type Number of Ports Used Notes
1 10 Gigabit Ethernet 4 Integrated 10GbE Ethernet controller

Network Configuration Notes

One 10GbE network interface from each 4100 node was connected to a Hitachi Apresia 15000-32XL-PSR switch, which provided network connectivity to the clients. The interfaces were configured to use Jumbo frames.

Benchmark Network

Each LG has an Intel XF SR10GbE single port PCIe network interface. Each LG connects via a single 10GbE connection to the ports on the Hitachi Apresia 15000-32XL-PSR switch.

Processing Elements

Item No Qty Type Description Processing Function
1 4 FPGA Altera Stratix IV EP4SE530 Filesystem
2 12 FPGA Altera Stratix IV EP4SGX360 Network Interface, Storage Interface, NFS
3 4 CPU Intel Xeon Quad-Core CPU Management
4 2 CPU Intel Xeon 8-Core CPU HUS VM I/O Management
5 2 ASIC Hitachi Custom ASIC HUS VM data engine

Processing Element Notes

Each 4100 node has 4 FPGAs that are used for processing functions. The HUS VM storage system is equipped with two Intel Xeon 8-Core CPUs and Hitachi custom ASICs.

Memory

Description Size in GB Number of Instances Total GB Nonvolatile
Server Main Memory 32 4 128 V
Server Filesystem and Storage Cache 68 4 272 V
Server Battery-backed NVRAM 8 4 32 NV
Cache Memory Module(HUS VM) 16 16 256 NV
Grand Total Memory Gigabytes     688  

Memory Notes

Each 4100 node has 32GB of main memory that is used for the operating system and in support of the FPGA functions. 68GB of memory is dedicated to filesystem metadata, sector cache and for other purposes. A separate, integrated battery-backed NVRAM module (8GB) on the filesystem board is used to provide stable storage for writes that have not yet been written to disk. The HUS VM storage system was configured with 256GB Memory.

Stable Storage

The Hitachi Unified Storage File node writes to the battery based (72 hours) NVRAM internal to the Server first. The data from NVRAM is then written to the backend storage system at the earliest opportunity, but always within a few seconds of arrival in the NVRAM. In a four node active-active cluster configuration, the contents of the NVRAM are synchronously mirrored (in a round robin fashion) to ensure that in the event of one or two node failover, any pending transactions can be completed by the remaining nodes. The data from the node is then written onto the battery backed backend storage system cache (a second layer of NVRAM in the entire solution) and are backed up onto the Cache Flash Memory modules in the event of a power outage. The Cache Flash Memory modules in the backend storage system are part of the total solution, but are used only during power outage and never used as cache space.

System Under Test Configuration Notes

The system under test consisted of four Hitachi Unified Storage File 4100 nodes, connected to a HUS VM All Flash storage system via two Brocade 5320 FC switches. The nodes were configured in an active-active cluster mode and are connected by a redundant pair of 10GbE connections to the cluster interconnect ports via two Brocade TurboIron 24X 10GbE switches (cluster interconnect switches). The HUS VM All Flash storage system consisted of 64 1.6TB Hitachi Accelerated Flash module drives. All the connectivity from server to the storage was via two 8Gbps switched FC fabric. For these tests, there were 2 zones created on each FC switch. Each 4100 server was connected to each zone via 2 integrated 8Gbps FC ports (corresponding to 2 FC ports). The HUS VM storage system was connected to the 2 zones (corresponding to 16 FC ports) providing I/O path from the server to storage. The System Management Unit (SMU) is part of the total system solution, but is used for management purposes only and was not active during the test.

Other System Notes

None

Test Environment Bill of Materials

Item No Qty Vendor Model/Name Description
1 16 Oracle Sun Fire x2200 RHEL 5 clients, two Dual core processors, 8GB RAM
2 1 Hitachi Apresia Hitachi Apresia 15000-32XL-PSR 10GbE Switch

Load Generators

LG Type Name LG1
BOM Item # 1
Processor Name AMD Opteron
Processor Speed 2.6 GHz
Number of Processors (chips) 2
Number of Cores/Chip 2
Memory Size 8 GB
Operating System Red Hat Enterprise Linux 5, 2.6.18-8.e15 kernel
Network Type 1 x Intel XF SR PCIe 10GbE

Load Generator (LG) Configuration

Benchmark Parameters

Network Attached Storage Type NFS V3
Number of Load Generators 16
Number of Processes per LG 200
Biod Max Read Setting 2
Biod Max Write Setting 2
Block Size 64

Testbed Configuration

LG No LG Type Network Target Filesystems Notes
62..77 LG1 1 /w/d0, /w/d1, /w/d2, /w/d3, /w/d4, /w/d5, /w/d6, /w/d7 None

Load Generator Configuration Notes

All the target filesystems from each node were accessed by all the clients.

Uniform Access Rule Compliance

All the filesystems from each node were mounted on all the clients. Each load generating client hosted 200 processes, accessing all the 8 target file systems (/w/d0, /w/d1, /w/d2, /w/d3, /w/d4, /w/d5, /w/d6, /w/d7).

Other Notes

Other test notes: None

Hitachi Unified Storage, Hitachi Unified Storage VM, Hitachi NAS Platform and Virtual Storage Platform are registered trademarks of Hitachi Data Systems, Inc. in the United States, other countries, or both. All other trademarks belong to their respective owners and should be treated as such.

Config Diagrams


Generated on Wed Oct 02 17:00:30 2013 by SPECsfs2008 HTML Formatter
Copyright © 1997-2008 Standard Performance Evaluation Corporation

First published at SPEC.org on 02-Oct-2013