dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
22014
share rss forum feed


sm5w2
Premium
join:2004-10-13
St Thomas, ON

File-copy transfer rate (wireless-N vs ethernet)

I'm copying files between a win-98se PC (desktop machine) and a windoze-7 netbook (HP mini-note 2140). I'm putzing around with a new wifi router (some cheapo Chinese unit - "netis" WF-2402) that I picked up recently for $20.

According to both the netbook and the router, I'm connected in 11n mode, with a connection speed of 135 Mbps. Distance between them for this test is about 20 feet.

I copy some music folder (about 100 mb) and time it with a stop watch. I get an effective data transfer rate of 317,300 bytes/second. Yes - 317 kbytes/sec.

I then connect the netbook directly to the Netis router and disable the wifi connection on the netbook. I repeat the test - and get 3,686,300 bytes/second.

With the netbook connected again via wifi, if I perform some internet speed tests, I get between 5.5 and 6.0 mbps download speed (my dsl sync is 7040/800).

So - given that, what sort of file-transfer speed should I get with a theoretical wifi connection of 135 Mbps?

Does it make sense that on wifi, my internet connectivity is 5.5/6 mbps while my file-transfer rate is effectively 317 kbytes/sec?


billaustin
they call me Mr. Bill
Premium,MVM
join:2001-10-13
North Las Vegas, NV
kudos:5
Ethernet is normally full-duplex, meaning it can transmit and receive at the same time. Wireless is half-duplex, it can either transmit or receive. Wireless (WiFi) also listens before sending, to make sure it is not interfering with another radio. Under ideal conditions, your wireless transfer rate will be just under half of the reported sync rate. Any other active wireless devices, that are within range, will cause the actual transfer rate to be even lower than expected.


sm5w2
Premium
join:2004-10-13
St Thomas, ON
> Ethernet is normally full-duplex, meaning it can transmit and receive at the same time.
> Wireless is half-duplex, it can either transmit or receive.
> Under ideal conditions, your wireless transfer rate will be just under half of the reported sync rate.

So if my wifi connection is listed as 135 mbps, then half of that is roughly 65 mbps. Assuming 10 bits per byte, that should work out to 6.5 M-bytes/sec. My actual file-transfer rate is *20 times slower* than that, at about 300 k-bytes/sec, while my apparent internet bandwidth test speeds are 6-mbps (600 k-bytes/sec).

> Wireless (WiFi) also listens before sending, to make sure it is not interfering with another radio.

There are no other wifi devices within 200 feet of this router. The router is "seeing" about a dozen other routers in the local area, the strongest coming in at 80% signal strength, the others at between 50 and 30% signal strength.

> Any other active wireless devices, that are within range, will cause the actual
> transfer rate to be even lower than expected.

How would you explain that internet bandwidth tests give 600k-bytes/sec download speed, while file-transfer rate across the local lan gives 300k-bytes/sec effective transfer rate?

Are there any wifi settings (packet size, mtu, etc) that have an impact on file transfer speed - but that don't otherwise impact internet TCP/IP packet transfer rate?


billaustin
they call me Mr. Bill
Premium,MVM
join:2001-10-13
North Las Vegas, NV
kudos:5

1 recommendation

There are not any settings you can change that will make things better. You could line your walls with metal, to eliminate outside interference, but that is an extreme step to take.

Make sure that you are comparing things correctly. Find a large file on the internet (at least 100Mb, ISO files are good for this). Download the file using a wired and wireless connection to the router with the netbook, and see if the download speed is the same. They should be very similar, the wired download may be a few seconds faster. If there is a significant difference, you have wireless issues (could be interference, could be the router).

Now, use the same file you downloaded. Transfer it from the netbook to the desktop and back again using wired and wireless connections. The wireless transfers should be about the same each direction. The wired transfers should be exactly the same each direction. The wired transfers should be significantly faster than the wireless. The wireless transfers should be significantly faster than the internet download.

The assumption is that the desktop PC is wired to the router, and the netbook is the only wireless device. The timed tests using the same file for all transfers will help to determine where any issues may be. The desktop PC could be a limiting factor, due to age or other issues. Wireless interference could be an issue. It may also be that you have a poorly functioning router. Try the exact same tests with another router and compare the results.


sivran
Seamonkey's back
Premium
join:2003-09-15
Irving, TX
kudos:1
reply to sm5w2
On my 802.11G, I never saw transfer rates above 2.5-3MB/s using a couple of laptops no more than 3 years old. Same story on a friend's wireless, also G.

My ancient proto-Netbook, with its 600MHz CPU and painfully slow hard drive, has never even broke 1MB/s with a wireless G PCMCIA card installed.

All numbers measured by FileZilla FTP and TeraCopy, connecting to a WRT-54GS running Tomato and.. whatever it is that AT&T U-Verse provided for my friend.

So, wireless is just slow, but yours is TOO slow! I'd expect a modern netbook to at least match what a three year old laptop can do, even on a cheapo Chinese piece of junk. speaking of which, have you tried using a different router? Take the netbook to a friend's place or something. Hell, go to a coffee shop. If you go faster on a cofee shop's free wifi, you'll know for sure your router's a piece of shit.
--
Think Outside the Fox.

HELLFIRE
Premium
join:2009-11-25
kudos:18

1 recommendation

reply to sm5w2
said by sm5w2:

what sort of file-transfer speed should I get with a theoretical wifi connection of 135 Mbps?

I've found a good rule of thumb for ANY IT claims is take the marketing number, and divide by 1/2 for the actual realworld
performance. Divide by 3/4 if the company's REALLY crappy.

One other test you could use, and to eliminate the HDD as a limiting factor, is get a program called IPERF, and run your
speed tests again.

Regards


captylor

@bellsouth.net
reply to sm5w2
did not say what wifi mode the router is in WPA2-PSK?
with the right setting for full N speed?
set in 20MHz mode or 40MHz?
MTU Setting 1492 for DSL?
With the slow transfer speed.
Does not seem like the settings are right for N or even G speeds.

With WiFi G best I have is 2.5-3MB/s max.
The Broadcom 4322AGN card in that netbook should be faster.


sivran
Seamonkey's back
Premium
join:2003-09-15
Irving, TX
kudos:1
reply to HELLFIRE
That would have to be one hella slow drive if his hdd was the bottleneck.


Da Geek Kid

join:2003-10-11
::1
kudos:1
reply to sm5w2
sorry did not read thru all the posts, but for higher thruput you may look into Wifi clients and APs using 2x2, 3x3 or even 4x4 MIMO to really give you 300+ Mbps.

In an enterprise class AP/ highend client I have seen 100+ thruput.

cramer
Premium
join:2007-04-10
Raleigh, NC
kudos:9
reply to sm5w2
His problem is wireless interference. There are too many neighboring wireless systems. An increase to 40MHz won't make any difference as the device will simply not do it if it see ANYTHING in the expanded range. Either switch to an 802.11a (5GHz) radio, or use a wired connection.

(In my old apartment, I used wired for everything as there were just too many wireless APs all over the place. This was not helped by every ISP installing wireless routers -- even when it wasn't used, the wifi was still on.)


Da Geek Kid

join:2003-10-11
::1
kudos:1
Reviews:
·Callcentric
thx. I did at least try to look for either 2.4Ghz or 5Ghz, but since I did not see anything and someone mentioned 40Mhz, I automagically assumed 5Ghz N using dual channel 40Mhz MIMO.

But yeah it is hopeless on consumer grade 2.4Ghz in an apt building filled with consumer grade 2.4Ghz. Recommendation: Run wires or buy 5Ghz AP and migrate over.


PapaDos
Cum Grano Salis
Premium,MVM
join:2001-02-08
Lasalle, QC
kudos:3
reply to sm5w2
Something must be very wrong, your WIRED speed is very bad.
You should fix the wired connection first...
--
Festina Lente


sm5w2
Premium
join:2004-10-13
St Thomas, ON
reply to sm5w2
Based on what I've read, the absolute theoretical max data rate on 100BASE-TX ethernet is 12.5 m-bytes/sec, and with protocol overhead that brings it down to 8 m-bytes/sec.

In my network, the primary router is a Linksys BEFSR81 (circa 2004). The desktop computer and Netis wifi router are connected to this Linksys router. So when I perform file-transfer tests between the desktop and the netbook, the data is passing through 2 routers.

I have the Netis connected in what I think of as "bridge mode" - ie the Netis is connected to the Linksys through one of it's 4 LAN ports - with no connection to the WAN port. I have assigned a fixed IP to the Netis, and turned off it's DHCP server. Any devices connected to the Netis get DHCP service from the Linksys.

So when I connect the netbook directly to one of the LAN ports on the Netis, my observed file-transfer speed of 3.7 m-bytes/sec is about half of what I think is the practical max of 8 m-bytes/sec.


Da Geek Kid

join:2003-10-11
::1
kudos:1
Reviews:
·Callcentric
huh???? 12.5*8 =100mbps. Theoretical? That's the speed that can be achieved. 8*8=64mpbs that's a little over half that. So not sure what kind of cables/hub you have but on a switch in a LAN setting between 2 PCs you could use jperf to get your real value.

oh, haha you maybe routing your traffic thru 2 different mediocre devices. Ok. not much to see here.

HELLFIRE
Premium
join:2009-11-25
kudos:18
reply to sivran
Agreed sivran See Profile, but in my line of work "if I can't measure / test it, it doesn't prove jack."

Regards