NetBench 7.0.3 Performance Report

 
 

"A Global Value-Added Distributor of new, used and refurbished computer periphery by a family-owned and operated company since 1979"
 Site Map

Product Categories

Battery BackUPS
CD/DVD Duplicators
Compaq
FibreChannel/HBA's
Hard Drives
Hard Drive Duplicators
HP
IBM
Iomega
Laptops
Networking Solutions
Media
Optical Drives
Processor/Video Upgrades
Printers
SCSI Adaptors
SUN
Tape Drives
UMAX Scanners
USB/FIREWIRE Solutions

c2it makes it easy to send and receive money online with your friends and family. Click here to see how:



PayPal lets any business or consumer with an email address to securely, conveniently, and cost-effectively send and receive payments online.

  NetBench 7.0.3 Performance Report
for INOSTOR® ValuNAS™ RAID 5 and
RAIDn 9:3
Summary
 

INOSTOR’s Testing and Quality Assurance group used Ziff-Davis Media’s NetBench®
version 7.0.3 to measure the throughput and response time of the INOSTOR® ValuNAS™
Network Attached Storage (NAS) appliance running RAID 5 and patented RAIDn 9:3.
ValuNAS introduces the use of serial ATA drives to the INOSTOR NAS product line.
 

RAIDn is a feature recently introduced to the data storage market. It produces parity data
that allows for the complete restoration of all lost data after the simultaneous failure of up
to three (any three) disks. This report describes the test procedures, documents the test
results, and compares those results with NetBench test results for the Dell 725N, as
published on the Internet.
 

RAID 5 is normally selected as the standard option where NetBench test results are used
in comparing different products. Running tests with both RAID 5 and RAIDn 9:3 not
only allows us to compare the performance of RAID 5 to RAIDn 9:3 on the ValuNAS, it
also enables us to compare the performance of ValuNAS running RAID 5 to the
published performance of the Dell 725N running RAID 5.
 

In both the RAID 5 and RAIDn 9:3 tests, INOSTOR’s ValuNAS, running with a gigabit
Ethernet card, exceeded industry norms for NAS products of its type in both throughput
rates and response times. Considering the extra parity calculations required for insurance
against the loss of up to three disks, we would not expect RAIDn performance to equal
RAID 5 performance. The actual throughput decline ranged between 0 and 10 percent,
based on the number of clients.

 

Running RAID 5, ValuNAS throughput measured over 314.9Mb/sec peak at 4 clients,
exceeding published Dell 725N 4-client performance by roughly 140% and exceeding
Dell’s performance for all valid client load levels used in the test.
 

The NetBench Test
 

Ziff-Davis Media’s NetBench version 7.0.3 is a portable PC Magazine benchmark
program that measures file server performance using a number of 32-bit clients making
network file I/O requests. Upon completion of the test suite, the NetBench program
provides an overall I/O throughput score and average response time for the tested device
and individual scores for the clients. The NetBench 7.0.3 Enterprise Disk Mix test
calculates response time in milliseconds and throughput in Mbits/seconds. The resulting
scores provide an industry-standard method for comparing relative performance that,
according to Ziff-Davis Media, are comparable to scores produced by other versions in
the NetBench 7.0.x series as well. NetBench can work with client PCs running any
combination of Windows 95/98, Windows 2000, Windows XP, or Windows NT/2000.
 

Test Bed Setup
 

For both the RAID 5 and RAIDn 9:3 testing, INOSTOR’s Testing group used the NetBench
Enterprise Disk Mix test suite to generate the throughput and response time values for a
9-drive ValuNAS 9000. All test-bed factors were as described below for both sets of
tests.
 

INOSTOR attached the ValuNAS 9000 to a network test bed in its lab consisting of 16 Dell
Dimension 4550 PC network clients, with each network client running on a Windows
2000 Service Pack 3 operating system. We connected each client to a GS524T Netgear
24-port gigabit switch. All ports on the gigabit Switch were set to run at 1000 Mbits per
second/full duplex. The connections from the ValuNAS to the network test bed used a
copper Gigabit port connection to the Netgear gigabit switch.
INOSTOR used the same basic setup for its NetBench controller; i.e., a Dell Dimension
4500 PC running Microsoft’s Windows 2000 Server (Service Pack 3). The controller
connected to the Netgear gigabit switch using an Intel PRO/1000 Mbits/second Desktop
network interface card (NIC) using a Cat 5E connection. The ValuNAS 9000 unit also
used a single copper Gigabit NIC connection during testing.
The 20.4GB Seagate Model ST320011A drives in the Dimension 4500 PCs have 2Mb
buffers, 16383 Cylinders, 16 Heads, and support an access time of 9.0ms or less.
 

To build the RAID 5 configuration, INOSTOR used the QuickStart setup wizard, a
standard feature of the software interface INOSTOR supplies with all of its current NAS
products. QuickStart prompted for a name to give the share and presented three options
relative to domain control: We could link to an external network domain controller, use
the ValuNAS itself as a domain controller, or operate with no domain controller. We
delegated domain controller responsibility to the same Windows 2000 Active Directory
server described above as the NetBench controller.
 

After we responded to the QuickStart prompts, the QuickStart program automatically
created a RAID 5 volume and an SMB/CIFS-accessible share without any additional
input on our part. We then mapped the share used by the test bed clients.
 

Beyond that, no additional tuning or configuration changes were required for system setup or share
access. We rebooted the ValuNAS, the network clients, and the controller before each
test run.
 

To build the RAIDn configuration, we used the full setup options of the iceNAS software
(INOSTOR‘s GUI). We configured a volume by selecting the RAIDn 9:3 option from a
pull-down menu. This selection requires that all available disks in the ValuNAS 9000 be
used in creating the RAIDn 9:3 volume. After the volume was partially constructed, we
were able to set up a share and enable it for SMB/CIFS. Again, we had three options for
domain control: link to an external network domain controller, use the ValuNAS itself as
a domain controller, or operate with no domain controller. We delegated domain
controller responsibility to the same Windows 2000 Active Directory server used in the
previous test.
 

We then mapped the share used by the test bed clients and tested to ensure that the share
was accessible. We rebooted the ValuNAS, the network clients, and the controller before
each test run.
 

Network Test Bed Clients
Machine Type
BIOS
Processors
L2 cache
Expansion Bus
Memory
Disks
Network Adapters
OS
OS updates
Dell Dimension 4550 PC
A03
Single 2.0 GHz Pentium ® 4
512K
400MHz
256 Mb
Seagate Model. ST320011A. 20.4GB. 2Mb buffer
Intel PRO/1000 Mbits/second Desktop NIC
 

Microsoft Windows 2000 Professional
Service Pack 3
NetBench Controller
Machine Type
BIOS
Processors
L2 cache
Expansion Bus
Memory
Disks
Network Adapters
OS
OS updates
Dell Dimension 4550 PC
A03
Single 2.0 GHz Pentium ® 4
512K
400MHz
256 Mb
Seagate Model. ST320011A. 20.4GB. 2Mb buffer
Intel PRO/1000 Mbits/second Desktop NIC
Microsoft Windows 2000 Professional
Service Pack 3
Network Configuration
Switches
Clients
GS524T Netgear 24-port gigabit switch


 

Test results
Table 1: NetBench Summary
C:\NetBench\Controller\Suites\NetBench\ent_dm.tst
Mix Name
Engines
Participating
Total
Throughput
(Mbits/sec)
Average
Response
Time
(milliseconds) Test Information
dm_1_clients 1 104.070 0.149 Engine Types: file
dm_4_clients 4 308.780 0.202 NetBench 7.0.3
dm_8_clients 8 297.946 0.425 Start Suite: Mon Dec 15 20:35:25 2003
dm_12_clients 12 260.411 0.733 Finish Suite: Mon Dec 15 23:39:32 2003
dm_16_clients 16 229.861 1.109 Elapsed Time: 03:04:07
dm_20_clients 16 230.842 1.106 Status: Suite completed successfully
dm_24_clients 16 228.887 1.117 Comments: valunas 9x3 build gigabit
dm_28_clients 16 229.699 1.110
dm_32_clients 16 228.867 1.116
dm_36_clients 16 227.975 1.120
dm_40_clients 16 227.094 1.126
dm_44_clients 16 229.298 1.115
dm_48_clients 16 228.317 1.120
dm_52_clients 16 229.675 1.112
dm_56_clients 16 231.095 1.106
dm_60_clients 16 229.071 1.117
 

Figure 1. INOSTOR ValuNAS 9000 NetBench 7.0.3 Throughput and Response Time – RAIDn 9:3 Configuration
 

Table 1: NetBench Summary
C:\NetBench\Controller\Suites\NetBench\ent_dm.tst
Mix Name
Engines
Participating
Total
Throughput
(Mbits/sec)
Average
Response
Time
(milliseconds) Test Information
dm_1_clients 1 104.014 0.149 Engine Types: file
dm_4_clients 4 314.669 0.199 NetBench 7.0.3
dm_8_clients 8 309.570 0.409 Start Suite: Mon Dec 15 08:34:41 2003
dm_12_clients 12 276.494 0.691 Finish Suite: Mon Dec 15 11:38:38 2003
dm_16_clients 16 255.757 0.998 Elapsed Time: 03:03:57
dm_20_clients 16 258.584 0.988 Status: Suite completed successfully
dm_24_clients 16 256.897 0.995 Comments: Raid 5 running gigabit valunas
dm_28_clients 16 257.625 0.992
dm_32_clients 16 258.152 0.990
dm_36_clients 16 257.521 0.993
dm_40_clients 16 257.140 0.994
dm_44_clients 16 256.299 0.998
dm_48_clients 16 256.785 0.995
dm_52_clients 16 259.725 0.983
dm_56_clients 16 259.737 0.984
dm_60_clients 16 258.046 0.990
 

Figure 2. INOSTOR ValuNAS 9000 NetBench 7.0.3 Throughput and Response Time – RAID 5 Configuration

 

 
     
© Copyright 2003.
*Price and availabilities subject to change without any notice. Not responsible for typographical errors.


 

tape drives | hard drives | optical drives | storage media | right storage solution for you |
 drive liquidation/specials

Archive | ADIC | Benchmark | BreeceHill |Certance|Compaq | Cybernetics| Dell | Exabyte | HP | IBM | Lacie| Maxoptix | Overland | OnStream | Plasmon | QualStar | Quantum | Quantum/ATL | Seagate | Sony | Spectralogic | Sun | StorageTek | Tandberg | TechMar | Wangtek

Drive FAQs: AIT Drives | LTO Drives | SDLT Drives | DLT Drives | LTO 2 VS SDLT 320