Jump to content
RESET Forums (homeservershow.com)

Real World Expectations about Data Transfer Speeds for N54L


Just Checking
 Share

Recommended Posts

In response to one of my posts on the USB 3.0 card topic, I asked about the transfer speeds I was seeing on my N54L.   Ikon suggested that I start a new thread and ask the general forum about their experiences.

 

My N54L Setup:

- N54L

- Mod-Bios to increase all port speeds to sata II, ACHI on all ports, etc.

- ICYDock dual 2.5" & 3.5" drive caddy.

- Vertex3 sata III 120Gb SSD as the OS & Program Drive on Port 5 with sata III cable. (In the ICYDock)

- Seagate 256Gb HDD (came with server) as the program support on port 4 (esata port) [in the ICYDock]

- 2 - WD Red 3Tb 7200rpm 64Mb cache HDD's running in a mirror array on ports 0 & 1 for Primary Storage.

- 2 - Seagate Barracuda 3Tb 7200rpm 64Mb cache HDD's running in a mirror array on ports 2 & 3

- Windows 7-64 bit Professional OS

 

I loaded the WD mirror drive with a mixed group of files totalling about 80Gb.  I then started copying that file back and forth to the Seagate mirror drive, putting the files in a different folder each iteration until I reached a total transfer size of about 650Gb. 

 

I watched the instantaneous transfer rate during the transfers and also looked at the average transfer rate (I'll have to get one of those programs which take the measurements and graph them).    My results showed that the peak transfer rate between the two mirror drives WD <=> Seagate was 77.2MB/sec.  The average transfer rate was around 51MB/sec.

 

This seems very slow to me.   When I transfer files between two of these HDD's in external enclosures, I see speeds of 100MB/sec on average.   Is the fact that I am using mirrored drives cutting down the transfer speed that much?  This is the first time I have ever tried to used mirrored drives.   I am just trying to set realistic benchmarks for this system.  

Link to comment
Share on other sites

I think one of your first tests would be to break the mirrors, then try the same copy operations between 1 WD and 1 Seagate. The idea, of course, is to establish if the mirror is actually affecting performance. Be sure to use the same drives, connected to the same ports; change as little as possible.

 

Now that I think about it, you could do 1 more test: break just 1 mirror and try the copy operation before breaking the other mirror.

 

I do have to say, though, I haven't experienced major performance issues when using mirrors. Typically, read operations are actually faster. However, it is possible the write operations are implicated.

Link to comment
Share on other sites

In response to one of my posts on the USB 3.0 card topic, I asked about the transfer speeds I was seeing on my N54L.   Ikon suggested that I start a new thread and ask the general forum about their experiences.

 

My N54L Setup:

- N54L

- Mod-Bios to increase all port speeds to sata II, ACHI on all ports, etc.

- ICYDock dual 2.5" & 3.5" drive caddy.

- Vertex3 sata III 120Gb SSD as the OS & Program Drive on Port 5 with sata III cable. (In the ICYDock)

- Seagate 256Gb HDD (came with server) as the program support on port 4 (esata port) [in the ICYDock]

- 2 - WD Red 3Tb 7200rpm 64Mb cache HDD's running in a mirror array on ports 0 & 1 for Primary Storage.

- 2 - Seagate Barracuda 3Tb 7200rpm 64Mb cache HDD's running in a mirror array on ports 2 & 3

- Windows 7-64 bit Professional OS

 

I loaded the WD mirror drive with a mixed group of files totalling about 80Gb.  I then started copying that file back and forth to the Seagate mirror drive, putting the files in a different folder each iteration until I reached a total transfer size of about 650Gb. 

 

I watched the instantaneous transfer rate during the transfers and also looked at the average transfer rate (I'll have to get one of those programs which take the measurements and graph them).    My results showed that the peak transfer rate between the two mirror drives WD <=> Seagate was 77.2MB/sec.  The average transfer rate was around 51MB/sec.

 

This seems very slow to me.   When I transfer files between two of these HDD's in external enclosures, I see speeds of 100MB/sec on average.   Is the fact that I am using mirrored drives cutting down the transfer speed that much?  This is the first time I have ever tried to used mirrored drives.   I am just trying to set realistic benchmarks for this system.  

 

 

Here are some benchmarks to give you a ballpark perspective though NOTE that the NASPT's aren't directly comparable between different test setups

 

clip_image027.jpg

Link to comment
Share on other sites

Thanks for the reference to your testing J_M.   I read your report and I will have to study it more thoroughly to be able to understand all the ramifications. 

Link to comment
Share on other sites

if you look at Joe_Miners benchmark tests you will notice that sequential file transfers show very high transfer speeds compared to directory transfers or files transfers in the benchmark.  A large single file sequential transfer is very easy for a computer and hard drives to accomplish.  When you get into many directories and smaller file sizes you increase the overhead of the transfers and it will slow down the system.  You increase the complexity to some extent with the mirrored drives.  I can't say exactly what you will see for a difference (yet).

 

Joe_Miner and I have been messing around with NASPT to see if we can figure out some of this stuff for benchmarks.

 

Your test was a little unusual because it was from drive set to drive set in the same box.  I don't think we can test that with the NASPT setup.

 

I want to also throw out there that you have different brand drives you are comparing (WD vs Seagate).  This can have an effect on the transfer rates as some drives are faster at reading than others so your results may vary depending on WD to Seagate or Seagate to WD.

Link to comment
Share on other sites

i wanted to do a more precise analysis of the performance testing i did, but was afaid it would take forever so i'm just posting some 'observations':

 

 

N54L internal 2 SSDs, copy speed, 200-220MBps between them. I like to do this test on new machines. I'm not sure of the specs for the two SSD, but they are new average SSDs. Just wanted to see the upper bound.

 

N54L internal two 72SB, single drives, 60-80 MBps.

 

N54L internal four 72SB,  2 stripes of two drives, using built in RAID,  180-190MBps.

 

N54L internal four 72SB,  2 stripes of two drives, using windows 7x64 stripe,  180-185MBps.

 

this is my production config:

N54L internal four ?Rpm Western Digital 3TB Green drives,  2 strips of two drives using windows 7x64 stripe, 175-185MBps copy speeds between them.

 

 

workstation w/ SSD - N54L prd config,   100-105MBps.  

 

workstation with 72SB - N54L prd config,  60-90MBps,  I am investigating the wide range.

 

the N54L prd config has two network cards. i can get both cards running at 100-105MBps at the same time while copying to two different workstations.

 

 

 

 

 

note 72SB= 1GB 7200rpm Seagate Barracuda.

speeds measured by doing a standard windows copy of large video files and eyeballing where the windows copy details speed settled at.
Link to comment
Share on other sites

if you look at Joe_Miners benchmark tests you will notice that sequential file transfers show very high transfer speeds compared to directory transfers or files transfers in the benchmark.  A large single file sequential transfer is very easy for a computer and hard drives to accomplish.  When you get into many directories and smaller file sizes you increase the overhead of the transfers and it will slow down the system.  You increase the complexity to some extent with the mirrored drives.  I can't say exactly what you will see for a difference (yet).

 

Joe_Miner and I have been messing around with NASPT to see if we can figure out some of this stuff for benchmarks.

 

Your test was a little unusual because it was from drive set to drive set in the same box.  I don't think we can test that with the NASPT setup.

 

I want to also throw out there that you have different brand drives you are comparing (WD vs Seagate).  This can have an effect on the transfer rates as some drives are faster at reading than others so your results may vary depending on WD to Seagate or Seagate to WD.

 

I just installed a SYBA PCIe 3 external port + 1 internal port USB 3.0 card in my N54L.   This allowed me to look at some numbers to transfer between external HDD's and the internal HDD's.  I have just downloaded some of the free software programs like CrystalMark but haven't had a chance to istall and try them.  I also got an inline power meter to look at the power usage of various components but that is for another post.

 

As CSKenney stated "A large single file sequential transfer is very easy for a computer and hard drives to accomplish.  When you get into many directories and smaller file sizes you increase the overhead of the transfers and it will slow down the system".   Yes, I found this to be very true.

That is why I used the same set of mixed files (actually, a backup of the files on one of my workstations) to take the measurements.  Those files consisted of MS Office documents and files, JPEG, Bitmaps, MPEG, WMA, MP-3, MP-4, FLV, AVI, program files, and iso files.  Actual file sizes ranged from 164 bits up to over 40GB.   This is a real world mix of the data that I want to back-up on the network server.

 

The results that I had previously shown indicated a peak transfer rate of around 77MB/sec and an average transfer rate of aroud 51MB/sec between the two mirrored drive sets.

 

I wanted to see if my USB 3.0 card was performing reasonably.   I connected up a Seagate 4TB Backup HDD with USB 3.0 to one port.   I connected a Seagate 2TB Backup HDD (7200 rpm) with USB 3.0 to another port.  Both of these drives came in their own enclosures.  I connected a Toshiba 2TB 7200rpm 64MB cache drive in a Vantec Nexstar3 USB 3.0 enclosure to the third external port on the SYBA card.  I started transferring the files back and forth between these drives and the mirrored drive and got the following results.

 

WD Red 3TB Mirror => Seagate 4TB USB 3.0 Backup - Peak: 52MB/sec;  Av.: 32MB/sec

WD Red 3TB Mirror => Seagate 2TB USB 3.0 Backup - Peak: 65MB/sec;  Av:  45MB/sec

 

At this point, I was ready to say that the SYBA card was worthless since the card was supposed to be able to transfer at up to 5Gb/sec speed (625MB/sec).   J_M had stated in a previous post that HDD's max out around 140-150MB/sec so I knew that I couldn't expect close to that.  Still, this was even less than between the two mirror drive sets which are supposed to be sata II or 3Gb/sec speed - 60% of the speed of the USB 3.0 card.  

 

I was getting bored waiting to have a 100GB file transfer which takes awhile to run a single test so I decided to just try the transfer of just one file.  I chose an iso file of a blueray disk (Skyfall) that I had in my library.  The file size was an almost exact 30GB.   These are the numbers that I measured:

Toshiba 2TB USB 3.0 => Seagate 4TB USB 3.0  - Peak: 147MB/sec;  Av: 142MB/sec.

Toshiba 2TB USB 3.0 => Seagate 2TB USB 3.0  - Peak: 200MB/sec;  Av: 145MB/sec.

Seagate 2TB USB 3.0 => WD Red 3TB Mirror  - Peak: 140MB/sec;  Av: 120MB/sec

Seagate 2TB USB 3.0 => Seagate 3TB Mirror  - Peak: 210MB/sec;  Av: 152MB/sec.

Seagate 3TB Mirror => Toshiba 2TB USB 3.0  -  Peak: 165MB/sec;  Av: 145MB/sec

 

These are speeds that are more like what I was hoping for!  As you can see, there is a difference between drives.  I believe that the only one that is significant is the speed to write to the WD Red Mirror drives.   That was significantly slower.   I'd have to repeat the test many times to state that it was statistically significant but those WD drives do seem slower than Seagate drives in general. 

 

I purchased the WD Red HDD's on the advice of my brother who likes those drives.  They cost me US$30 more per drive than the equivalent size Seagate Barracuda HDD's.   I have never had a Seagate Barracuda drive fail on me.   The current reviews of the WD Reds when I bought those drives was that there were high failure rates.   Several people on this forum have stated they had WD Reds fail.   The slower transfer rates on these WD Red drives makes me even less happy that I spent 25% more money for them than the equivalent Seagate Barracuda HDD's.

 

There appears to be no difference between transfer rates between drives on the USB3.0 card and transferring data to and from the card to the drives in the server.  I attribute this to the limitation of the HDD's, not to the limitations of the SYBA PCIe USB 3.0 card.   I will have to connect in another SSD to the internal USB 3.0 port on the SYBA card and do more tests with that.  

 

I was originally unhappy with the Toshiba drive when I first got it because it kept clicking (armature) during read/writes and it seemed to run hotter than my Seagate Barracud drives.   After I ran the Toshiba drive for 24 hours straight it queted down and now doesn't make any more noise than the Seagate drives or run hotter.  I am still worried that the drive will fail on me prematurely.  I shall see.   It does seem to be as fast as the equivalent size Seagate Barracuda HDD. 

Link to comment
Share on other sites

It appears that the failure rate of Seagate drives is highly dependent on the specific model: some models seem to be very reliable while others seem prone to quite a lot of failure. That's the only way I can account for the "I've never had a Seagate drive fail." versus the "I'll never buy another Seagate drive." statements.

 

I would expect the WD Reds to be somewhat slower, particularly at writing, than the Seagate. The Seagate drive is 7200 RPM; the Reds are ~5400 RPM. The Reds are really designed to be used in groups, as in a NAS arrangement: i.e. RAID. One other difference is that the Red drives should be significantly more power efficient than the Seagate. They do run with almost zero vibration and almost at ambient temperature.

Link to comment
Share on other sites

Because I'm trying to get some real world numbers to benchmark what I can expect, I tried some sample transfers between the HDD's attached to the server and some workstations.   These are laptops connected wireless to the network.   I use two D-Link DIR-825 Dual Band N600 WiFi routers (to get double the bandwidth in the 2.4GHz band) for my network.   The routers are hooked up in series from the cable modem and the server is connected directly to the primary WiFi router through the gigabit ethernet port.   I run the latest version of firmware on the D-link routers (I don't use dd-wrt or open wrt) but I have modified them with dual 9db High Gain antennas.  I get very high signal strength of excellent quality in a dead stable configuration.  

 

I did a transfer of the iso file (30GB) from the internal Seagate 3TB Mirror drive to a laptop that was 10ft away from the WiFi router antenna (to ensure that I got maximum high quality signal strength)  my laptop (HP Elitebook 8460p) measured signal strength of -19db which is as good as it gets.   The transfer speeds that I measured were  Max: 7.6MB/sec; Av: 6.1MB/sec.   If I want to do backups of the drives on the workstations it would take over 4 hours to back-up just the files at that rate. 

 

Doing multiple tasks on the WiFi network will slow things down further.   I did a test where I streamed video from an external drive on the server (Toshiba 2TB USB 3.0) to a laptop (Elitebook 8460p) on the 2.4GHz band of the secondary router and also streamed another video from the same external drive to another laptop (HP Probook 4525s) on the 2.4GHz band of the other primary WiFi router.  This worked perfectly even when I moved the laptops more than 15 meters away from the routers and put 5 walls and lots of furniture between them.   While Streaming to the two laptops, I started simultaneously streaming Youtube videos on a couple smartphones connected to the network.   Again, this was successful.   

 

As a final test, I transferred an iso file from one laptop through the network to a HDD on the server using a second laptop to control the transfer while simultaneously streaming back a video signal from the external HDD on the server to the laptop which was sending the iso file to the server.   I measured a file transfer rate of 850KB/sec maximum and an average of 650KB/sec.   

 

I tried transferring the mixed data files I usually use wirelessly from the 4525s laptop to the server with nothing else on the network.  The rates I saw were Max: 3.2MB/sec and Av: 1.6MB/sec.   At that rate it would take around 20 hours to back up the files on one of these laptops.   I would not even venture to guess how long it would take to do a full clone of the HDD for a bare metal backup.

 

I believe that my wireless network is as fast as I can expect.   It isn't a question of bandwidth of the WiFi routers since I have two of them and the transfer rate I measured doesn't even come close to the capacity of the single 2.4GHz band of one of the routers (I do have QoS enabled) let alone the capacity of two routers on 4 bands.  The Elitebook has an internal dual band N wireless card supposedly capable of 300 Mb/sec on each band.   The Probook 4525s has an internal single band n card supposedly capable of 300Mb/sec but I suspect 150Mb/sec. 

 

My questions to the forum are:

1. What kind of WiFi transfer rates are reasonable expectations?

2. What do other people see?

3. Is it possible to implement an automatic back-up of the laptop workstations with a wireless network?  (Do it at night?) 

 

I do have Hard Wire  (CAT5E) at the location.  I also have a switch to put in.   I just don't have the time to put it in and the scheduled priority to make that happen.   I need to make do for now with the wireless network.

I ran a cou

Link to comment
Share on other sites

Wireless backup of laptops to WHS2011 is very viable. You might want to use hardwire for the 1st one, since it will take the longest. I believe MS recommends doing restores over hardwire.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...