TS-219P II expected performance

[LukaszG]

Hi,

I just bought a TS-219P II array and I'm wondering what transfer rates I can realistically expect. Figures shown around the Internet vary from 30-100 MB/s for reads/writes. On my current setup I'm achieving about 60-65 MB/s and frankly I'm a bit disappointed, since I was always able to transfer files between two PCs over the same network at ~100 MB/s (and I'd expect the dedicated $400 drive array to be of no worse performance - or am I naive?).
I've updated the array's firmware. I'm using a 2 TB WD Caviar Green drive for a single drive config. I've read a lot here that these can cause troubles in RAID setups, but I'm guessing that should not be the cause of slower transfer rate (according to internet reviews, the drive is capable of achieving >100 MB/s when connected directly to a PC - I haven't verified that myself).

Any help will be appreciated.
Lukasz

[P3R]

On my current setup I'm achieving about 60-65 MB/s and frankly I'm a bit disappointed...

Sounds fairly decent to me but yes, according to specifications it should be able to do around 100 MB/sec. The latter is however most likely achieved in a pure lab environment with possibly performance tuned equipment, not in an ordinary home network.

For anybody to be able to provide any useful comments or suggestions on improvements you need to supply much more of the exact configurations of all affected equipment (NAS-network-client), protocol used, test procedure (only a large single file transfer is vaild), what tool is used for reporting and with separate read/write numbers.

...since I was always able to transfer files between two PCs over the same network at ~100 MB/s (and I'd expect the dedicated $400 drive array to be of no worse performance - or am I naive?).

The PCs likely have a fast Intel-processor that all by itself use much more power than the whole NAS (including disks) does. That's not really a fair comparison.

Qnaps are primarily very power-frugal, pre-packaged, relatively easy-to-manage appliance servers. They weren't originally intended to compete for the title of being the fastest servers on earth, even if the streamlining of the system do give them decent performance and the later top models (TS-X79) actually have very respectable specifications.

I'm using a 2 TB WD Caviar Green drive for a single drive config. I've read a lot here that these can cause troubles in RAID setups, but I'm guessing that should not be the cause of slower transfer rate (according to internet reviews, the drive is capable of achieving >100 MB/s when connected directly to a PC - I haven't verified that myself).

According to this Storagereview.com test, a WD Caviar Green perform at best at 116.3/111.8 MB/sec sequential read/write but not more than 53.8/52.27 MB/sec at worst. Are you sure the disk performed at it's absolute best (outer tracks) in your specific test?

[LukaszG]

P3R thank you for taking time to respond.
Indeed I don't know which part of the platter was involved in the test (I just copied a 4 GB file with Total Commander), so it's possible that the file landed on the slower side - (although, since the drive is virtually empty, I'd assume that the firmware utilises the faster regions of the drive first?).
As for my setup, as I said I was copying a large file with Total Commander from a HP 8710w laptop (Windows 7, Core2Duo CPU, 8 GB RAM, Samsung SSD) via a 1 Gb/s router (I've seen 100+ MB/s transfer rates from this laptop to my other machine on regular basis so I know that, network-wise, the setup is capable of such speeds). Today, I also used CrystalDiskMark v.3.0.1, which, for the sequential 1 GB transfer test reported 64 MB/s read and 74 MB/s write speeds.
Unless you have any other recommendations, I am going to buy a 3 TB WD Red drive for this NAS after the weekend. When I'll have it, and I'll do more tests on it (I'll try to do sequential tests in different regions of the platter to whether it's a factor). I have to deal with large files quite often in my work, so even a 20% bump in transfer rate can translate into non-negligible time savings.

[LukaszG]

P3R thank you for taking time to respond.
Indeed I don't know which part of the platter was involved in the test (I just copied a 4 GB file with Total Commander), so it's possible that the file landed on the slower side - (although, since the drive is virtually empty, I'd assume that the firmware utilises the faster regions of the drive first?).
As for my setup, as I said I was copying a large file with Total Commander from a HP 8710w laptop (Windows 7, Core2Duo CPU, 8 GB RAM, Samsung SSD) via a 1 Gb/s router (I've seen 100+ MB/s transfer rates from this laptop to my other machine on regular basis so I know that, network-wise, the setup is capable of such speeds). Today, I also used CrystalDiskMark v.3.0.1, which, for the sequential 1 GB transfer test reported 64 MB/s read and 74 MB/s write speeds.
Unless you have any other recommendations, I am going to buy a 3 TB WD Red drive for this NAS after the weekend. When I'll have it, and I'll do more tests on it (I'll try to do sequential tests in different regions of the platter to whether it's a factor). I have to deal with large files quite often in my work, so even a 20% bump in transfer rate can translate into non-negligible time savings.

[P3R]

Indeed I don't know which part of the platter was involved in the test (I just copied a 4 GB file with Total Commander), so it's possible that the file landed on the slower side...

I only added that as one of the things that is easy to miss and that may severly affect performance. My point is that performance testing isn't trivial, and it requires a lot of very careful work to give reliable results.

- (although, since the drive is virtually empty, I'd assume that the firmware utilises the faster regions of the drive first?).

I assume the same but you didn't reveal the fill level previously, so nobody but you keow that.

As for my setup, as I said...

As far as I can see, you haven'ty told the forum any of the details below previously.

...I was copying a large file with Total Commander from a HP 8710w laptop (Windows 7, Core2Duo CPU, 8 GB RAM, Samsung SSD) via a 1 Gb/s router (I've seen 100+ MB/s transfer rates from this laptop to my other machine on regular basis so I know that, network-wise, the setup is capable of such speeds).

That's half of the testing environment so what about the other half, the NAS network configuration and cables used to connect it.

[LukaszG]

Indeed I don't know which part of the platter was involved in the test (I just copied a 4 GB file with Total Commander), so it's possible that the file landed on the slower side...

I only added that as one of the things that is easy to miss and that may severly affect performance. My point is that performance testing isn't trivial, and it requires a lot of very careful work to give reliable results.

- (although, since the drive is virtually empty, I'd assume that the firmware utilises the faster regions of the drive first?).

I assume the same but you didn't reveal the fill level previously, so nobody but you keow that.

As for my setup, as I said...

As far as I can see, you haven'ty told the forum any of the details below previously.

The "as I said" was actually referring to the previous paragraph in the same post (and not to my original post) and was a just stylistic ornament (and definitely not a way to point out any omissions or anything like that).

...I was copying a large file with Total Commander from a HP 8710w laptop (Windows 7, Core2Duo CPU, 8 GB RAM, Samsung SSD) via a 1 Gb/s router (I've seen 100+ MB/s transfer rates from this laptop to my other machine on regular basis so I know that, network-wise, the setup is capable of such speeds).

That's half of the testing environment so what about the other half, the NAS network configuration and cables used to connect it.

The NAS network config is pretty mundate - I use the default settings: speed auto negotiation (currently at 1000 Mbps) and 1500 MTU. I use the Ethernet cable provided with the NAS.

[P3R]

The "as I said" was actually referring to the previous paragraph in the same post (and not to my original post) and was a just stylistic ornament (and definitely not a way to point out any omissions or anything like that).

I'm sorry, I misunderstood that.

The NAS network config is pretty mundate - I use the default settings: speed auto negotiation (currently at 1000 Mbps) and 1500 MTU. I use the Ethernet cable provided with the NAS.

Check for any network errors on the System Status, Sustem Information page (last column).

If that is okay and you're sure nothing else is a problem with your hardware I guess you'll have to find the bottleneck somewhere in the software configuration.

It's a fact that most users don't see the maximum performance specified by Qnap and many come here to complain about that. Personally I find it hard to beleive that Qnap would deliberately fake their specifications and I believe they somehow did achieve the numbers they claim. It's probably a needle-in-the-haystack-task to find the thing that is the culprit. I'm sorry but I don't think I have the time to help you find it. Good luck!