Currently I run a Dlink DIR-625 router with a 20/5 fios connection coming into it and routing to my desktop via wired, 2 laptops via wifi, and my brothers computer via wifi. I get about 12-16mbps max download speed and ~6mbps cross network transfers. Even 3" from the router with 100% signal strength. I've tried other routers as well. Is this an inherent issue with wireless due to network overhead and signal interference or is my network broke?
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
802.11G network not even nearing 54mbps? Is this normal?
- Thread starter z3r0-
- Start date
Shut down all computers (easiest way to ensure they aren't transmitting) other than your wired desktop and laptop and try a file transfer between the two.
Then try a file transfer on the internet somewhere.
See what you end up getting.
Sounds like it might be set to B rather than G to me... Check your settings.
And just how do you come to that conclusion? Rather than posing something ELSE I was going to post- I'll let you explain yourself first (because I am rather curious myself how he has "no chance" of this).
Well if he/she has tried different routes then it is not with the router and 16mb is somewhat acceptable for a 11g network anyway.
If you can find anyone who can get over 20mb on non pre-n or mimo routers then I would like to speak to them.
Hmm interesting... My laptop uses the Intel PROSet 3945ABG card and sometimes drops the connection. I found the fix for this is to roll back to MS drivers instead of Intel. I did that and, without any change in network load, my DL speeds hit 19mbps steady. That's acceptable.
bobdole369
Gawd
- Joined
- Jun 27, 2004
- Messages
- 856
When dealing with data transfer, you have the line rate (in your case 54mbps), and then there is the actual data rate. Add collisions, failed acks, duplicate acks, encryption, latency, etc, and essentially what you are seeing is correct.
Take 54mbps (thats mega-BITS per second), you would think that in a perfect world you divide 54mbps by 8 (since there are 8 bits in a byte) = 6.75 megaBYTES per second. Nope - you add a start bit and stop bit, so you divide by 10. 54 mbps = 5.4MBps, or 5400 Kbps
TCP packets are sent, must be acked, then the ack must be acked. This adds latency. Factor in the environment (no data over teh air is 100%, I don't care if your right next to the router, your going to lose some packets, and these must be retransmitted.)
TCP overhead is usually not factored into data throughput tests, definitely not if you tx a file and time it with stopwatch. It is about 10% of the data sent over the network.
802.11g is half duplex. All wireless stuff is. Reduce your data rate by another 20-40% because of this. Again each packet must be acked, and the ack acked.
Typical wired 100mbit full duplex throughput is about 60--70 mbits, or around 6-7 megabytes a second. I've seen 30mbit/second (3 megabytes a second) over a heavily loaded 802.11g (about 25 users doing database stuff) with WPA-AES on an enterprise grade WAP. This is about as good as it gets.
Since your cross network transfers are with 2 wireless connected hosts - divide your throughput in half again, as you double the number of acks and add more packet loss/latency.
i.e. you send a packet to the router, its acked by the router, your computer acks the ack. The router sends the packet to your brothers computer - your brothers computer acks the packet, the router acks the ack. Like I said, even 3" from the router your going to get packet loss. It might be one out of a thousand packets, but it will happen. When it does, the whole thing stops until the packet is retransmitted and re-acked and re-ack acked. This really slows things down. Half or less of your typical transfer rate is usually the case when you have 2 wireless clients working with each other, so I would say your throughput is about 16 mbps. Not that shabby on 802.11g. (b networks usually have actual throughput of about 1-3mbps).
I would see if you somehow have a B client in there, which would activate "mixed mode" or "cts/rts protection", and slows the network considerably. The B client only has to be associated to slow it down. Your router might have an option for "G-Only".
It's working well, but it might not be optimal.
Take 54mbps (thats mega-BITS per second), you would think that in a perfect world you divide 54mbps by 8 (since there are 8 bits in a byte) = 6.75 megaBYTES per second. Nope - you add a start bit and stop bit, so you divide by 10. 54 mbps = 5.4MBps, or 5400 Kbps
TCP packets are sent, must be acked, then the ack must be acked. This adds latency. Factor in the environment (no data over teh air is 100%, I don't care if your right next to the router, your going to lose some packets, and these must be retransmitted.)
TCP overhead is usually not factored into data throughput tests, definitely not if you tx a file and time it with stopwatch. It is about 10% of the data sent over the network.
802.11g is half duplex. All wireless stuff is. Reduce your data rate by another 20-40% because of this. Again each packet must be acked, and the ack acked.
Typical wired 100mbit full duplex throughput is about 60--70 mbits, or around 6-7 megabytes a second. I've seen 30mbit/second (3 megabytes a second) over a heavily loaded 802.11g (about 25 users doing database stuff) with WPA-AES on an enterprise grade WAP. This is about as good as it gets.
Since your cross network transfers are with 2 wireless connected hosts - divide your throughput in half again, as you double the number of acks and add more packet loss/latency.
i.e. you send a packet to the router, its acked by the router, your computer acks the ack. The router sends the packet to your brothers computer - your brothers computer acks the packet, the router acks the ack. Like I said, even 3" from the router your going to get packet loss. It might be one out of a thousand packets, but it will happen. When it does, the whole thing stops until the packet is retransmitted and re-acked and re-ack acked. This really slows things down. Half or less of your typical transfer rate is usually the case when you have 2 wireless clients working with each other, so I would say your throughput is about 16 mbps. Not that shabby on 802.11g. (b networks usually have actual throughput of about 1-3mbps).
I would see if you somehow have a B client in there, which would activate "mixed mode" or "cts/rts protection", and slows the network considerably. The B client only has to be associated to slow it down. Your router might have an option for "G-Only".
It's working well, but it might not be optimal.
Did you already optimize the connection on your Windows XP PC at Verizon's web site? If not, do so now and reboot:
http://www2.verizon.net/help/fios_settings/optimizer/
At 54Mbps 802.11g wireless sync, 21-22Mbps is usable and the remaining 33-34Mbps is used for overhead.
Most newer D-Link routers have a Status -> Wireless screen that will tell you what speed your wireless devices are synced at. The 16Mbps figure suggest a sync speed of 48Mbps, while the 12Mbps would suggest a sync speed of 36Mbps or less.
As noted by others, 2.4GHz cordless phones and 2.4GHz (Bluetooth) wireless keyboards and mice can interfere with your wireless signal.
The D-Link Status -> Statistics page shows the number of wireless errors. Are your wireless errors incrementing regularly? If so, that is a key indication of interference.
Note if you live in Florida, it is currently 'normal' to see less throughput than the 20Mbps advertised, because the backbone used by Verizon in Atlanta (which your traffic has to travel through) is congested. Verizon expects to upgrade those links later this year, but for now, FiOS customers in Florida will only see 10-15Mbps sustained throughput. If you live anywhere else, you should see the full advertised bandwidth on FiOS.
http://www2.verizon.net/help/fios_settings/optimizer/
At 54Mbps 802.11g wireless sync, 21-22Mbps is usable and the remaining 33-34Mbps is used for overhead.
Most newer D-Link routers have a Status -> Wireless screen that will tell you what speed your wireless devices are synced at. The 16Mbps figure suggest a sync speed of 48Mbps, while the 12Mbps would suggest a sync speed of 36Mbps or less.
As noted by others, 2.4GHz cordless phones and 2.4GHz (Bluetooth) wireless keyboards and mice can interfere with your wireless signal.
The D-Link Status -> Statistics page shows the number of wireless errors. Are your wireless errors incrementing regularly? If so, that is a key indication of interference.
Note if you live in Florida, it is currently 'normal' to see less throughput than the 20Mbps advertised, because the backbone used by Verizon in Atlanta (which your traffic has to travel through) is congested. Verizon expects to upgrade those links later this year, but for now, FiOS customers in Florida will only see 10-15Mbps sustained throughput. If you live anywhere else, you should see the full advertised bandwidth on FiOS.
MIRTBelGeran
Weaksauce
- Joined
- Apr 28, 2006
- Messages
- 126
Another thing to look at is do you have any other devices that run @ 2.4GHz in your house. like wirless phones. Get rid of the old 2.4 phones and go with the 5Ghz Sometimes you can clear up isseus by chnging chanels
BobSutan
[H]F Junkie
- Joined
- Apr 5, 2000
- Messages
- 12,121
There are many things that can negatively impact your wireless network. Some things to look for: microwaves, cordless phones, and other access points in use.
Fire up netstumbler and see if anyone else is using the same channel as your AP. If so, change it by 5 channels (i.e. 1, 6, 11 are customary channels to use in densely populated areas).
Fire up netstumbler and see if anyone else is using the same channel as your AP. If so, change it by 5 channels (i.e. 1, 6, 11 are customary channels to use in densely populated areas).