NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Struggling With Low Speeds < 15MBps Over Gigabit
Hey Guys
I have been having a major PITA regarding transfer speeds ever since I purchased my 'ReadyNas Ultra 4 Plus' in May 2011.
Originally, I was only getting approx 5-6 MB/second and came to the conclusion that it was probably due to using 'Netgear Homeplugs', which apparently didn't support jumbo frames even though they were the 500MBps version .
I also discovered that the integrated Broadcom network adapter within my 'Dell XPS8300' didn't support jumbo frames either, despite that being Gigabit. :shock:
Requiring faster transfer speeds, I ditched the 'Netgear Homeplugs' and invested in an 'Intel Gigabit CT' network adapter as well as a TP-Link TL-SG1005D5 5-port unmanaged Gigabit switch.
This increased the speed to approx 10-12 MB/second and I've put up with that for a while now.
However after recently reading a webpage, 'Bott's Thoughts' dated 2010, I was disappointed to read that I should be expecting around 80 MB/second :shock: .
So I want to try and get to the bottom of this poor network performance once and for all.
Current set up consists of:
'ReadyNas Ultra 4 Plus'
RAIDiator 4.2.23 (v4.2.24 Available)
4 x 2TB SAMSUNG HD204UI - 5400RPM - RAID-5
4GB DDR2 Memory Upgrade
Auto Negotiation @ 1Gbit/Full Duplex
Jumbo Frames/Packets Enabled
IPv6 Disabled
Main PC
'DELL XPS8300'
Intel i7-2600 CPU @ 3.4GHz
Window 7 64-bit
2 x 1TB Western Digital WD10EALX - 7200RPM RAID-0
16GB DDR3
Disabled Onboard Broadcom Gigabit Network Adapter (No Jumbo Frame/Packet Support)
Intel Gigabit CT Network Adapter
Auto Negotiation @ 1Gbit/Full Duplex
Jumbo Frames/Packets Enabled = 9014
IPv6 Disabled
Router/Switch
Billion BiPAC 7800N
Firmware 1.06h
4 x Gigabit Ports
Jumbo Frames Enable by Default (Apparently)
DHCP Server
IPv6 Disabled
Unmanaged Switch
TP-Link - TL-SG1005D
5 x Auto-Negotiation Gigabit Ports
Belkin CAT5e Cables / All Tested
Currently, the ''Dell XPS8300' PC is connected directly to the 'Billion 7800N' router/switch (both upstairs), which is then connected to the unmanaged switch downstairs via CAT5e cable.
The switch is then connected to:
'ReadyNas Ultra 4 Plus'
MEDE8ER Multimedia Streamer
'SKY HD' Receiver
XBOX360 Slim
Samsung BD-C5900 3D Bluray Player
For test purposes:
I copied a 3.2GB file from the 'ReadyNAS' to the 'Dell XPS8300' and the average Windows reported read speed was approx 100MBps
I copied the same file back and the average Windows reported write speed was 11MB/second. :?
Moved the same file from the 'ReadyNAS' to the 'Dell XPS8300' and the average Windows reported read speed was approx 120MBps
Moved the same file back and the average Windows reported write speed was 10MB/second. :shock:
Done the same with a 'HP 8000 Elite USDT', which is also Gigabit and the results were similar.
The results from a utility called 'LAN Speed Test' (3GB test file to the same folder) show the following:
Average Read Speed: 29MB/second
Average Write Speed: 18MB/second
Surely the write speed to the 'ReadyNAS' should be better than 10-12MB/second?
The only thing I haven't tried is a direct cable connection between the 'Dell XPS8300' and the 'ReadyNAS' as I don't have a cable long enough as present.
Going to order a Belkin 20m CAT5e or CAT6 cable now and see if that improves speeds any.
Is there anything obvious I've missed?
Looking at the above, I'm now wondering whether the difference in drive speeds between the 'Dell XPS8300' and the 'ReadyNAS' is causing this.
If so, would the expected transfer rate be affected that much or is it more likely to be an external hardware or even a software/driver problem?
Any ideas welcome.
Oh, and if you got this far without losing the will to live, then well done! You deserve a medal! :D
I have been having a major PITA regarding transfer speeds ever since I purchased my 'ReadyNas Ultra 4 Plus' in May 2011.
Originally, I was only getting approx 5-6 MB/second and came to the conclusion that it was probably due to using 'Netgear Homeplugs', which apparently didn't support jumbo frames even though they were the 500MBps version .
I also discovered that the integrated Broadcom network adapter within my 'Dell XPS8300' didn't support jumbo frames either, despite that being Gigabit. :shock:
Requiring faster transfer speeds, I ditched the 'Netgear Homeplugs' and invested in an 'Intel Gigabit CT' network adapter as well as a TP-Link TL-SG1005D5 5-port unmanaged Gigabit switch.
This increased the speed to approx 10-12 MB/second and I've put up with that for a while now.
However after recently reading a webpage, 'Bott's Thoughts' dated 2010, I was disappointed to read that I should be expecting around 80 MB/second :shock: .
So I want to try and get to the bottom of this poor network performance once and for all.
Current set up consists of:
'ReadyNas Ultra 4 Plus'
RAIDiator 4.2.23 (v4.2.24 Available)
4 x 2TB SAMSUNG HD204UI - 5400RPM - RAID-5
4GB DDR2 Memory Upgrade
Auto Negotiation @ 1Gbit/Full Duplex
Jumbo Frames/Packets Enabled
IPv6 Disabled
Main PC
'DELL XPS8300'
Intel i7-2600 CPU @ 3.4GHz
Window 7 64-bit
2 x 1TB Western Digital WD10EALX - 7200RPM RAID-0
16GB DDR3
Disabled Onboard Broadcom Gigabit Network Adapter (No Jumbo Frame/Packet Support)
Intel Gigabit CT Network Adapter
Auto Negotiation @ 1Gbit/Full Duplex
Jumbo Frames/Packets Enabled = 9014
IPv6 Disabled
Router/Switch
Billion BiPAC 7800N
Firmware 1.06h
4 x Gigabit Ports
Jumbo Frames Enable by Default (Apparently)
DHCP Server
IPv6 Disabled
Unmanaged Switch
TP-Link - TL-SG1005D
5 x Auto-Negotiation Gigabit Ports
Belkin CAT5e Cables / All Tested
Currently, the ''Dell XPS8300' PC is connected directly to the 'Billion 7800N' router/switch (both upstairs), which is then connected to the unmanaged switch downstairs via CAT5e cable.
The switch is then connected to:
'ReadyNas Ultra 4 Plus'
MEDE8ER Multimedia Streamer
'SKY HD' Receiver
XBOX360 Slim
Samsung BD-C5900 3D Bluray Player
For test purposes:
I copied a 3.2GB file from the 'ReadyNAS' to the 'Dell XPS8300' and the average Windows reported read speed was approx 100MBps
I copied the same file back and the average Windows reported write speed was 11MB/second. :?
Moved the same file from the 'ReadyNAS' to the 'Dell XPS8300' and the average Windows reported read speed was approx 120MBps
Moved the same file back and the average Windows reported write speed was 10MB/second. :shock:
Done the same with a 'HP 8000 Elite USDT', which is also Gigabit and the results were similar.
The results from a utility called 'LAN Speed Test' (3GB test file to the same folder) show the following:
Average Read Speed: 29MB/second
Average Write Speed: 18MB/second
Surely the write speed to the 'ReadyNAS' should be better than 10-12MB/second?
The only thing I haven't tried is a direct cable connection between the 'Dell XPS8300' and the 'ReadyNAS' as I don't have a cable long enough as present.
Going to order a Belkin 20m CAT5e or CAT6 cable now and see if that improves speeds any.
Is there anything obvious I've missed?
Looking at the above, I'm now wondering whether the difference in drive speeds between the 'Dell XPS8300' and the 'ReadyNAS' is causing this.
If so, would the expected transfer rate be affected that much or is it more likely to be an external hardware or even a software/driver problem?
Any ideas welcome.
Oh, and if you got this far without losing the will to live, then well done! You deserve a medal! :D
36 Replies
Replies have been turned off for this discussion
- Under Status > Health in Frontview what are the SMART+ stats for your drives like?
- -In addition, please confirm that jumbo frames are actually working on your network.
You can do this with ping. On a PC go to the windows start bar, and enter "cmd" That will open a dos box.
Enterping NASIPADDRESS -f -l 8960
If you get "Packet needs to be fragmented but DF set" then lower the 8960 value until you find the largest value that works, and let us know what that is.
Jumbo frames don't always speed things up btw, so it would also be a good idea to test with jumbo frames turned off.
-You also might want to confirm that you are using cat 5e or cat 6 ethernet cable. The cable grade is usually printed along the length of the cable every now and then. mdgm wrote: Under Status > Health in Frontview what are the SMART+ stats for your drives like?
SMART+ stats seem OK and everything is green.
Disk 1 SAMSUNG HD204UI 1863 GB , 22 C / 71 F , Write-cache OFF OK
Disk 2 SAMSUNG HD204UI 1863 GB , 23 C / 73 F , Write-cache OFF OK
Disk 3 SAMSUNG HD204UI 1863 GB , 24 C / 75 F , Write-cache OFF OK
Disk 4 SAMSUNG HD204UI 1863 GB , 23 C / 73 F , Write-cache OFF OK
Temp CPU 50 C / 122 F [Normal 0-80 C / 32-176 F] OK
Temp SYS 29 C / 84 F [Normal 0-65 C / 32-149 F] OK
StephenB wrote: -In addition, please confirm that jumbo frames are actually working on your network.
I initially tried the 8960 value you mentioned and sure, I got the message 'Packet needs to be fragmented but DF set'.
I kept trying different values until I reached 1449.
It was at this point I got the following:Pinging 192.168.1.2 with 1449 bytes of data:
Reply from 192.168.1.2: bytes=1449 time<1ms TTL=64
Reply from 192.168.1.2: bytes=1449 time<1ms TTL=64
Reply from 192.168.1.2: bytes=1449 time<1ms TTL=64
Reply from 192.168.1.2: bytes=1449 time<1ms TTL=64
Ping statistics for 192.168.1.2:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 0ms, Average = 0ms
When I tried 1450 or anything higher as the value, I got the 'Packet needs to be fragmented but DF set' message again.
As for cables, all are at least CAT5e with the odd CAT6 in the mix.- The ping test shows that your network doesn't carry jumbo frames. In fact, you are not getting the normal ethernet MTU of 1500 - you start running into problems at 1490 or larger. (you add 40 bytes to the ping value you found to get the max frame size).
You did test this over ethernet correct? WiFi doesn't carry jumbo frames.
So I suggest turning jumbo frames off for now, and reduce the frame size to 1480, and retest. Packet fragmentation will certainly reduce throughput.
Then try to figure out what component is interfering with the jumbo frames. If all your devices are connected to the trendnet switch, then it is not router. If you have devices connected to the router,try moving them to the switch (as most consumer routers do NOT support jumbo frames).
On the disks, I'm not sure what the program fail cnt is about, but the other parameters look fine. The program fail cnt is Samsung specific, none of my disks source that parameter.
Well, that's a bit alarming.StephenB wrote: The ping test shows that your network doesn't carry jumbo frames. In fact, you are not getting the normal ethernet MTU of 1500
Yeah, done over ethernet.StephenB wrote: You did test this over ethernet correct? WiFi doesn't carry jumbo frames.
I will try that tomorrow.StephenB wrote: So I suggest turning jumbo frames off for now, and reduce the frame size to 1480, and retest. Packet fragmentation will certainly reduce throughput.
I'm gonna have to strip every thing to the bare essentials I think and keep adding items until I find the one that's causing the problem.Thank for the adviceStephenB, it's been very useful.StephenB wrote: Then try to figure out what component is interfering with the jumbo frames. If all your devices are connected to the trendnet switch, then it is not router. If you have devices connected to the router,try moving them to the switch (as most consumer routers do NOT support jumbo frames).
Me neither, but it does look worrying.StephenB wrote: On the disks, I'm not sure what the program fail cnt is about, but the other parameters look fine. StephenB wrote: ........try to figure out what component is interfering with the jumbo frames.
Well, I've stripped it back to basics.
'DELL XPS8300'
>
Belkin Cat6 UTP Patch Cable
>
'ReadyNas Ultra 4 Plus'
I actually had to join 2 x Cat6 cables joined with a brand new 'Startech Cat5e RJ45 Inline Coupler', because Amazon failed to deliver the 'Belkin 15m Cat6 Snagless STP Patch Cable' on time. :(
Anyhow, not good news!
Despite the connection being direct, the results from 'Lan Speed Test' and from Pinging were the same.
The result from 'Lan Speed Test' was a bit faster when I disabled jumbo frames at both ends (average of 23MBps), but pinging would would still crap out at 1449.
So, it's either the ReadyNAS or the PC.
While it's still set up like it is, I'll reboot the PC and re-enable the onboard LAN and try that.
At least then if it's the same result it's still either the ReadyNas or PC.
If the results are much faster, then I'll know it's the Intel 'Gigabit CT' card or associated software/settings that probably to blame.
Oh, one thing that I forgot to ask about speed expectation was, by what method of transfer are they calculated?
My speed results are based on 'Drag & Drop' via the Windows explorer interface. If this is not the best option, can some advise on a better method (both user friendly with fast transfer).
Again, thanks StephenB- You can also use NAStester 1.4, which will do a timed drag and drop for you. There's a download link here: http://www.808.dk/?code-csharp-nas-performance
It seems reasonably accurate.
Disabling jumbo frames of course would not change the ping results MTU test. It just tells the devices not to use jumbo frames.
Did you also reset the MTU to 1480 on the NAS and in your NIC settings? StephenB wrote: You can also use NAStester 1.4, which will do a timed drag and drop for you.
I'll check that out soon.StephenB wrote: Did you also reset the MTU to 1480 on the NAS and in your NIC settings?
Every time I change the Jumbo Frames setting on the NIC, I also do the same on the ReadyNAS. The MTU on the ReadyNAS is automatically altered between 1500-9000 when selecting/deselecting Jumbo Frames.- Well, I've done a few speed tests and the results can be seen below.
1.Desktop PC > ReadyNAS
Static IP
Cat6 Cable
2. Desktop PC > Unmanaged Gigabit Switch > ReadyNAS
Static IP
Cat6 Cable
3. Desktop PC > Billion 7800N Router/Switch > Unmanaged Gigabit Switch > ReadyNAS
DHCP
Cat5e + Cat6 (Mixed)
All other equipment was disconnected from router, switch etc during tests.
Not scientific by any means and the table is a little complicated:
Just tried that NASTester program and did a couple of runs (5 x 400MB + 1 x 2000MB) and the resulting write speed was approx 26MB/Sec
I'd be interested to see what speed others achieve or what other programs, file protocols, etc that others use to achieve better transfer speeds.
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
