Archive Home arrow Reviews: arrow Network arrow QNAP TS-259 Pro Turbo NAS Server

QNAP TS-259 Pro Turbo NAS Server E-mail
Reviews - Featured Reviews: Network
Written by Bruce Normann   
Monday, 03 May 2010
Table of Contents: Page Index
QNAP TS-259 Pro Turbo NAS Server
QNAP Turbo NAS Features
QNAP TS-259 Pro NAS Hardware
QNAP TS-259 Pro Software
QPKG Center Software Expansion
Closer Look: QNAP TS-259 Pro
Insider Details: QNAP TS-259 Pro
QNAP v3 User Interface
NAS Testing Methodology
Basic-Disk Test Results
Windows 7 Disk Test Results
NAS Server Final Thoughts
QNAP TS-259 Pro Conclusion

Windows 7 Disk Test Results

I started this article with my standard test system, which has been running Windows 7 since it launched last October. It's the 64-bit version because there's no reason for most users to continue the 32-bit legacy. It's the Ultimate version because that's what Microsoft was giving away at their media events. I don't really need the additional features of the Ultimate package, and all my other PCs in the house run Windows 7 Home Premium. I wasn't expecting any issues or problems; it's been a very stable platform for over six months now, and all I would be doing is transferring a few files back and forth to another device on the network. Boy was I wrong!

Everything worked alright; in fact it worked too well. When I compared the results I was getting with the results for other NAS devices reviewed here on Benchmark Reviews, I was stunned. I was getting read and write results 50-60% better than comparable units, that were tested quite recently with similar hardware. I retested, and then I tried to equalize things a bit by turning off some new features, like Remote Differential Compression. No matter how I tried to cripple Win7, it always produced significantly faster file transfers than previous tests with Windows XP.

At this point, I popped a new hard drive into the test bench system, did a fresh install of XP and then Service pack 3. Then I downloaded fresh drivers for the onboard NIC and started retesting. Well, that fixed it, so to the results were very much in line with prior tests. These are the numbers I reported above, where I compared the TS-259 with the other NAS servers. In those charts we are comparing apples-to-apples; in this section of the review let's look at the performance differences between Windows XP and Windows 7 on a single NAS server, as they are significant.


Looking at read tests, where we read a file from the NAS and copy it to a HDD on the test system, Win7 consistently trounces Windows XP. On average, the results are 53% quicker with Win7, and are pretty much the same whether copying a 1GB file or a 10GB file. Looking at these results, you can see why I was nonplussed at first when I started running the tests. The numbers are not even close, and the individual tests were very steady from run to run. I can't say I'm unhappy with the change in performance; who wouldn't like a 50% speed increase, especially on a task that generally takes several minutes to complete. Let's take a look at write performance.


The write tests, where I copied a file from the test system to the NAS, offer a very similar picture. Indeed, the average performance increase for Win7 is 53% again. In this case, the Jumbo Frame benefits with Win7 were more apparent, and helped pull the numbers up. Without Jumbo Frames, with an MTU of 1500, Win7 was only 45% faster than XP. With Jumbo Frames enabled, the performance increase jumps up to 60%. Anyone doing backups to a NAS device is going to be very pleased with this kind of improvement. Think about how long a nightly backup takes now, and knock off a third of that time. Same thing if you do full backups every week or every month; you can get it done in two thirds the time it takes now.

Most likely, you will be copying thousands of smaller files rather than one or two large ones, which will slow things down. So you won't achieve the same transfer rates as we did, but the improvements should be similar when you upgrade to Windows 7. For most users this is not enough justification for upgrading their operating system, but its one heck of a perk if you were going to do it anyway.



# Test with bonding gbit lan ?^-Super_Treje-^ 2010-05-03 23:34
No test with the network in "bonding" ?
Report Comment
# I did, but....BruceBruce 2010-05-04 07:15
I repeated the tests with IEEE 802.3ad Link Aggregation Control Protocol, using two Intel Gigabit CT Desktop Adapters in the test bench system. The problem with that test scenario and Teaming or Bonding or whatever you want to call it, is that the network speed stays exactly the same. The bandwidth is increased by widening the data path, not increasing the speed. I.e. it?s analogous to two fully loaded trucks driving the speed limit instead of one truck delivering your data. Yes, you get twice the data, but you get it in the same time frame, which is what our testing measures.

I think the way to test this feature is to have two or more transfers occurring at the same time. With one transfer already under way, another could be started and timed, and the speed of the second transfer should be relatively unaffected by the continued activity of the first one. Your thoughts, suggestions?
Report Comment
# 802.3ad is NOT your solutionscavenger 2012-11-30 12:01
YES this is it. Load balancing is made only on multiple file transfers.

If you can read french, I posted a lot about it on but the result is this one :
Conclusion is 802.3ad is ONLY failover. ABSOLUTELY NOT load balancing.
If you want to do what I dreamed of, choose on each side the Balance-SLB (or Balance-ALB) + round robin transmit load balancing method.
Then you will have a smooth repartition of the packets on each port, but you will notice a strong down bandwidth due to the fact that "Packet order is NOT guaranteed"
Load balancing for a one file transfer on many cables is just a dream... right now...
Report Comment

Comments have been disabled by the administrator.

Search Benchmark Reviews

Like Benchmark Reviews on FacebookFollow Benchmark Reviews on Twitter