Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[gunrunnerjohn]

I still have to believe this is the switch, nothing else makes much sense here!

[Hannes123]

on a local LAN where there is only one path between source & destination, out of sequence packets should not occur (assuming of course, that that is what is meant be "inconsistencies in the sequence numbers")

In the frames I captured the sequenz numbers somethimes did not make sense.
Everytime the sequenz numbers were not correct, this started with a frame going from the NAS to the Notebook, which was not numbered as expected. The notebook replied to such packages with the acknowledgement to the last package it received, that was correctly numbered. It did't this as often as necessary to force the retransmission of the package with the sequenznumber it expected .

I get equal results with Linux. The transfere rates to the streaming client I use are very low also.

I will contact my dealer and try to get the switch replaced. If the problems are caused by the switch, this will be the cheapest way to find out.

[fordem]

I have a question here which is not 100% related.

Are you running Wireshark on the notebook - in other words - the same system that you are transferring data from?

[DrDigital]

The first thing I would do is connect the DNS-323 directly to the computer in question and remove any sort of network/switch issue and see if your throughput is where you think it should be.  This can be done by using static IP on the DNS and your computer and connecting the two with a cable (you may have to use a null ethernet cable if either side doesn't automatically do xover).

[Hannes123]

he first thing I would do is connect the DNS-323 directly to the computer in question and remove any sort of network/switch issue

Instead of a direct connection between computer and NAS I used a second Gbit-switch to test the connection. It worked without any problems.

Are you running Wireshark on the notebook - in other words - the same system that you are transferring data from?

Yes!

[dvdsham]

Hi. I read a comment here on the forum about the difference between a cat 5e and a 6 -- the truth of it is, there is no difference. A standard straight through Ethernet cable and cross overs only use 4 wires (two wires for transmitting and two wires for receiving. Power Over Ethernet uses 6 wires and or all 8 wires for only when an electrical hardware on off switch is used to a POE device. All wires are in a twisted pair so that a constant 10 Mhz is being generated between devices.

If you have a Gigabit switch or router but your computer Network Adapter is only 10/100 then a Gigabit switch is not benefiting your speed. You need a 10/100/1000 NIC in your computer in order for a Gigabit switch or router to be beneficial.

[jamieburchell]

there is no difference

Cat 6 cables are physically different - better shielding etc, possibly more twists per inch. They are less susceptible to interference, allowing for greater throughput.

[gunrunnerjohn]

CAD5e is fully rated for gigabit, you won't see any difference in any normal environment.

[fordem]

CAD5e is fully rated for gigabit, you won't see any difference in any normal environment.

I'll assume that was a typo, and yes CAT5e is fully rated for gigabit, and you CAN in many cases run gigabit over CAT5, there IS a difference between CAT5e & CAT6 cables - as jamieburchell pointed out, the number of twists per inch IS different, causing better noise rejection.

I might have a graph somewhere from a Fluke DSP showing the difference between CAT5 & CAT6 cable - I'll see if I can find it, and post it.

[gunrunnerjohn]

It should be obvious that there are differences in the cable, or there wouldn't be a different type. 

However, if we get back to the topic of this thread, a good CAT5e would NOT be the cause of the issues described, perhaps instead of wandering through the weeds, we could actually address the topic at hand?

[MrBOFH]

I just ran a test based on everything above. My setup for that is as follows

Win 7 64bit Home Premium machine with onboard Realtek PCIe GBE GB ethernet LAN
Trendnet 8 port GB green switch TEG-S80G
DNS-323 with 2 Samsung Spinpoint F3 1GB drives in RAID1 (firmware 1.09)
Cables used are manufactured Cat 5e.

Both the Win 7 box and DNS-323 have  9KB MTU Jumbo frames enabled

512MB exe file copied from Windows to DNS323 - 19MB/s
512MB exe file copied from DNS323 to WIndows - 30MB/s

Before I actually ensured I had 9KB jumbo frames set on both the Windows box and the DNS, my throughput was more around 10-13MB/s. Both units are set to auto for link speed.

Same test but with an HP laptop (win 7 64bit Home Premium)  that only has a 100MB network controller.
Connected to a Trendnet 5 port GB green switch TEG-S50G which is directtly connected to the TEG-S80G that connects to the DNS-323.

Same 512MB file.
Laptop to DNS - 9.5MB/s
DNS to Laptop - 11.5MB

My above win 7 box to the laptop copy of the file was about 13MB/s so the lan controller on the laptop only being 100M is definately the bottleneck here.