Open-E's DSS V6: Storage Software Set Up, Managed, And Benchmarked

Test System And Benchmarks

Of course, we couldn’t help but to benchmark Thecus' appliance with Open-E. The N8800PRO is currently our only eight-bay NAS server, precluding comparisons to competing models. However, in the interest of including some degree of baseline performance, we do have data for a couple of five-bay servers, Qnap's TS-559 Pro+ and Synology's DS1511.

You'll want to take these results in context. Depending on your own target environment and performance-oriented settings (stuff like the block size, RAID algorithm, physical disks operating concurrently, or addition of the 10 Gb Ethernet option), the throughput you see could be higher, or even lower. There's also the fact that we're comparing different drives. The N8800PRO submitted to us included Hitachi Deskstar 5K3000 disks, while the drives installed in the Qnap TS-559 Pro+ and Synology DS1511 were Samsung HD103SJs.

Swipe to scroll horizontally
System Hardware
MotherboardAsus P5E3 Deluxe, Rev. 1.03G, Intel X38, BIOS: 0810 (11/02/2007)
CPUIntel Core 2 Duo E6750 (65 nm Conroe core) @2.26 GHz
RAM2 x 1024 MB Crucial Ballistix DDR3-1600
eSATA ControllerJMicron JMB363
System DiskSeagate Barracuda 7200.9, 160 GB, 7200 RPM, SATA 3Gb/s, 8 MB Cache
Data Disks8 x 3.5" Hitachi Deskstar 5K3000, 2000 GB, 5400 RPM, SATA 6Gb/s, 32 MB Cache
DVD ROM DriveSamsung SH-D163A , SATA150
Graphics CardGigabyte Radeon HD 3850 GV-RX385512H, GPU: 670 MHz, RAM: 512 MB DDR3 (830 MHz, 256-bit)
Network CardMarvell Yukon 88E8056 PCIe Gigabit Ethernet Controller
AudioIntegrated
Power SupplyCooler Master RS-850-EMBA, ATX 12V V2.2, 850 W
System Software & Drivers
Operating SystemWindows Vista Enterprise SP1
Graphics DriverATI Radeon Version 7.12
Network Driver9.0.32.3 (Vista-Standard)
Intel Chipset DriverVersion 6.9.1.1001 (20/02/2008)
JMicron Chipset DriverVersion 1.17.15.0 (24/03/2007)

Intel NAS Performance Toolkit

We use Intel's NAS Performance Toolkit to test the NAS devices that make their way through our lab. You can read more about this software in Benchmarking With Intel's NAS Toolkit.

Sound Level and Power Dissipation

It is fairly obvious that a rack-mounted server like the N8800PRO was never intended to live its life next to your workstation. Its power consumption, which hovers around 114 W, isn't too bad. However, its acoustic output is less tolerable. Eight total fans generate a sound level of 53.4 db(A). This is certainly no personal video server.

  • dgingeri
    I manage about 1200 servers for a software test lab of 9 departments. I work with many different types and generations of hardware. I can tell you from experience, I would absolutely HATE this machine, no matter how fast it performs. No console? Only network communication? Horrible idea. I've been dealing with a few NetApp units that were set up this way. They are a major HEADACHE to manage. If the information on the IP address for this machine were lost, it would be almost impossible to get it to work again. There are admins out there (like my predecessor at my current job) who don't do much for documentation. Small businesses typically don't even have their own admins, so imagine a business switching admin companies because of pricing debates, and a new guy comes in after a drive failure. Basically, he's have to tell them their nice storage system is useless because he has no idea how to get into it to tell it to rebuild to a new drive. They'd have to replace the whole thing.

    For the hardware price, I'd go for a Dell R510. it's about the same price. Then I'd run CentOS 5 or 6 on it. CentOS can do most of what this thing can do with no cost. Or the business could opt for Win 2k8 r2 for $600 more or MS SBS for $800 more and get an easier to manage system that any yahoo proclaiming to be an admin could fix.
    Reply
  • Argo16
    I totally agree with dgingeri and I would add that most of these products do not meet the quality standards required by business-class storage. They are substantially poorly built oversized SOHO appliances. I doubt that the DESY project and many health care providers are using this kind of storage.
    Reply
  • Why did not you include 5 drive test together with 8 drive test? How could we know it's Open-E being a TOTAL LOSER or it's just comparing apples (8 drives) to oranges (5 drives)?
    Reply
  • peter_b123
    I can tell TH that I use Open-E DSS v6 in some production envorionments for my SMB users and I've been very satisfied with its performance, configuration and support. Don't let one bad review fool you. I consistently get over 250MB/s with DSS systems that I've built myself.
    Reply
  • peter_b123
    I can tell TH that I use Open-E DSS v6 in some production envorionments for my SMB users and I've been very satisfied with its performance, configuration and support. Don't let one bad review fool you. I consistently get over 250MB/s with DSS systems that I've built myself.
    Reply
  • peter_b123
    I can tell TH that I use Open-E DSS v6 in some production envorionments for my SMB users and I've been very satisfied with its performance, configuration and support. Don't let one bad review fool you. I consistently get over 250MB/s with DSS systems that I've built myself.
    Reply
  • So what do we get for $1600? No drives? OK... Crappy case with non-working tiny LCD and a 2 year old desktop mobo with ancient CPU? Fine... And a crappy software from near dead Open-E? How lovely! This money can buy you HP or IBM or Dell server with Xeon CPU (probably with 2 socket option), tons of RAM, option for SAS and 8 drive bays. Go install OpenIndiana on it with Napp-it and ZFS thing is going to run circles around Open-E based old boy for just a fraction of cost! Good luck Thecus!!!
    Reply
  • TheKurrgan
    I've never been a fan of these types of devices...
    IOP performance is garbage on SATA drives, and I've seen FreeNAS do a better job than what was portrayed in this article.
    As far as the performance, I'd say it was right in line with low end devices using iSCSI.
    Any iSCSI devices you can pretty much count the 1GB ethernet link as the bottle neck, which puts it around 100 - 125 max. 2x Multi pathing may bring that up to around 190, but good luck achieving usable transfer rates of 250+ as peter claims.
    For the money i'd choose dgingeri's dell in a heart beat.
    Reply
  • cozsmin
    I had once openfiller on a vmware

    Al changes , done to the storage , were done in some xml files of the application

    Needles to say that whenever you did anything in command line , the web apllication would not recognize the result
    Reply
  • Slothy
    While the impression was already there that Tom's IT is just one big advertising centre, it seems there is a slider for Tom's Hardware reviews. The closer you get to IT and away from their core of consumer-grade hardware/software articles, the more the articles become an obvious marketing release and the less useful they become.

    That said, keep up the good work in your core space Tom's!
    Reply