NAS Benchmarking Database

Quick Links to Benchmark

Avatar Benchmark MP3 Benchmark 2GB Transfer Test Intel NASPT IOMeter (iSCSI) CrystalDiskMark Power Consumption

When we evaluate a NAS we do more than just benchmark it.  We look at every feature of the product in a real world environment.  The NAS becomes our main file server, media server, or whatever other functions it can perform.  Each aspect of the product is examined, from software to hardware to performance.  The NAS is connected to our network just as it would be to yours.

First, we connect the NAS directly to our switch using the Ethernet cables provided in the packaging.  If no Ethernet cables are provided, we will use 5ft CAT6 cables.  If the NAS features 3 or more drive bays, we will conduct our benchmarks using RAID 5.  Also, if the NAS has the capability of running various filesystems, we will conduct our benchmarks using ext4 if supported.  Subsequently, if the NAS has only 2 drive bays we will test the NAS in both RAID 1 and RAID 0.

If a NAS is provided with no drives, we will populate all drive bays with Hitachi 2TB consumer grade hard disk drives.

All benchmarks are run a minimum of 5 times and the results are averaged.  We reboot our bench system between each run and defragment our hard drive if necessary before running the benchmark.

Continue on to the next page to see our NAS Benchmark Comparison

Motherboard	Gigabyte 880GA-UD3H
Processor	AMD Phenom II X4 965
Memory	2x4GB Kingston HyperX DDR3
Hard Drive	Crucial M4 SSD 128GB
Chassis	Lian Li PC-6
Power Supply	BFG LS-450
Operating System	Windows 7 Professional 64-bit
NIC	2 x Intel Gigabit CT Adapters (Teaming – IEEE 802.3ad)
Network Switch	Netgear GS716Tv2

Avatar Benchmark

For testing the large file transfer speed of the NAS, we transfer a 46.6GB Blu-Ray ISO (Avatar) from the workstation to the NAS and back five times.  We pipe the transfer speeds during the benchmark to a .csv file for charting.  For more on our Avatar Benchmark please see our Benchmarking application: COD Benchmarking Utility

MP3 Benchmark

For testing the multimeda transfer speed of the NAS, we use the same method for the avatar test and create two hundred MP3’s and specific file sizes for repeatable results.  The files are copied from the workstation to the NAS and back 5 times.  We pipe the transfer speeds during the benchmark to a .csv file for charting.

Intel NASPT

For further benchmarking, we test each NAS using Intel’s NAS Performance Toolkit to rate of 2 key areas of NAS performance.  We measure File Copy speeds to and from the NAS and the Directory Copy speeds to and from the NAS.  System RAM is set to 2048MB of RAM to ensure there is no mem caching for transfers.

The Intel® NAS Performance Toolkit (NASPT) is a file system exerciser and analysis tool designed to enable direct measurement of home network attached storage (NAS) performance. Designed to emulate the behavior of an actual application, NASPT uses a set of real world workload traces gathered from typical digital home applications. Traces of high-definition video playback and recording, office productivity applications, video rendering/content creation and more provide a broad range of different application behaviors. With the latest version of NASPT, users may even add their own custom traces. NASPT reproduces the file system traffic recorded in these traces onto whatever storage solution the user provides, records the system response, and reports a rich variety of performance information.

http://software.intel.com/en-us/articles/intel-nas-performance-toolkit/

C.O.D. 2GB Transfer Test

For testing the large file transfer speed of the NAS, we transfer a 2GB file from the workstation to the NAS and back five times. We pipe the transfer speeds during the benchmark to a .csv file for charting.

IOMeter (iSCSI)

For benchmarking the performance of iSCIS, we created an iSCSI connection to the NAS and benchmark the NAS using Intel’s IOMeter.  IOMeter has to be one of the most widely used software packages for benchmarking disks. There are an endless number of ways you can build a benchmark, but we chose to limit our benchmarking to 100 percent Sequential Read and Write then 100 percent Random Read and Write. You can emulate our benchmarking by using the following configuration

‘size	% of size	% reads	% writes	% sequential	% random	delay	burst	align	reply
2,097,152	100	100	0	100	0	0	1	0	0
2,097,152	100	0	100	100	0	0	1	0	0
2,097,152	100	100	0	0	100	0	1	0	0
2,097,152	100	0	100	0	100	0	1	0	0

CrystalDiskMark

CrystalDiskMark is another benchmarking utility that determines speeds by measuring 512KB, 4KB, and 4KB (Queue Depth 32) sequential and random read and write speeds. Test data can be Random, 0Fill, and 1Fill. For our purposes, we left the test at Random.

Power Consumption

We conduct our power consumption using a Kill A Watt EZ P4460 Power Meter and measurements are taken every minute for 10 minutes then averaged for the results. During this benchmark, the NAS is stressed for heavy transfers, left for normal usage, then idle until we achieve low power mode or HDD standby.