I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little. I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get. Thanks.
Oliver Jones wrote:
I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little.
I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get.
Thanks.
_______________________________________________ wlug mailing list | wlug(a)list.waikato.ac.nz Unsubscribe: http://list.waikato.ac.nz/mailman/listinfo/wlug
That sounds about right. What you need to understand is that the 54Mbit data rate is the absolute maximum under laboratory conditions. By the time you add protocol overhead, hardware limitations, software limitations and RF propergation limitations the reality is 2Mbits is about the best stream rate there is. Under "burst" conditions you may get better then 2Mbits but that is assuming perfect RF conditions.
I hope you mean 2 MBytes there as I certainly get more than 5 % effiency! On 27/08/05, DrWho? <x_files_(a)ihug.co.nz> wrote:
Oliver Jones wrote:
I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little.
I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get.
Thanks.
_______________________________________________ wlug mailing list | wlug(a)list.waikato.ac.nz Unsubscribe: http://list.waikato.ac.nz/mailman/listinfo/wlug
That sounds about right.
What you need to understand is that the 54Mbit data rate is the absolute maximum under laboratory conditions.
By the time you add protocol overhead, hardware limitations, software limitations and RF propergation limitations the reality is 2Mbits is about the best stream rate there is.
Under "burst" conditions you may get better then 2Mbits but that is assuming perfect RF conditions.
_______________________________________________ wlug mailing list | wlug(a)list.waikato.ac.nz Unsubscribe: http://list.waikato.ac.nz/mailman/listinfo/wlug
On Sat, 2005-08-27 at 10:43 +1200, DrWho? wrote:
That sounds about right.
What you need to understand is that the 54Mbit data rate is the absolute maximum under laboratory conditions.
By the time you add protocol overhead, hardware limitations, software limitations and RF propergation limitations the reality is 2Mbits is about the best stream rate there is.
Under "burst" conditions you may get better then 2Mbits but that is assuming perfect RF conditions.
Can you provide some backing for these statements? They seem completely and utterly wrong to me. The theoretical maximum throughput of an 802.11g device in a clean environment close to the AP (no RF interference, etc) is around 30Mbps. This is a comparable level to the 6Mbps theoretical maximum throughput of 802.11b, both protocols have a theoretical maximum throughput of around 55% of the actual serialisation rate. Theories are nice, but I've never seen a 802.11b link operate at 6Mbps, I've seen 5.5Mbps though. Most of the g links I've seen have operated at about 6 Mbps, but they were in no way representative of a "normal" 802.11g link so I wouldn't take that figure as truth. The point I want to make is that DrWho's statements seem completely wrong, in ideal situations I would expect you to be able to achieve at least 10+ Mbps using 802.11g and that in my experience if you're only getting 2Mbps then there is something wrong. See http://www.proxim.com/learn/library/whitepapers/parameters_802.11g_performan... for some backup for the points and speeds I've quoted above. Cheers -- Matt Brown matt(a)mattb.net.nz Mob +64 275 611 544 www.mattb.net.nz
DrWho? wrote:
Oliver Jones wrote:
I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little.
I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get.
That sounds about right.
What you need to understand is that the 54Mbit data rate is the absolute maximum under laboratory conditions.
By the time you add protocol overhead, hardware limitations, software limitations and RF propergation limitations the reality is 2Mbits is about the best stream rate there is.
Under "burst" conditions you may get better then 2Mbits but that is assuming perfect RF conditions.
I get 280KB/s using konqueror/fish on my 802.11b. This link goes from my house to my garage/office and the link quality is not good because the garage has steel cladding. I have the access point taped to the window and I have line of sight to it from the kitchen table through an aluminium framed window. This is the only place in the house where it works reliably, even moving to the other side of the table causes it to have problems. Given that my setup is not ideal but I get more than twice your throughput on 802.11b I suspect you should be able to get better than 120Kb/s on your link. Here's what iwconfig says about my card. The link quality goes up and down between 40/92 and 70/92. The 280 Kb/s is what it gets when the link quality is around 60-70. /sbin/iwconfig lo no wireless extensions. eth0 no wireless extensions. eth1 IEEE 802.11-DS ESSID:"WLAN" Nickname:"flap" Mode:Managed Frequency:2.412 GHz Access Point: 00:08:A1:42:9D:BD Bit Rate:11 Mb/s Tx-Power=15 dBm Sensitivity:1/3 Retry min limit:8 RTS thr:off Fragment thr:off Power Management:off Link Quality=66/92 Signal level=-44 dBm Noise level=-148 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:10 Invalid misc:0 Missed beacon:0 -- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
Glenn Ramsey wrote:
I get 280KB/s using konqueror/fish on my 802.11b.
This link goes from my house to my garage/office and the link quality is not good because the garage has steel cladding. I have the access point taped to the window and I have line of sight to it from the kitchen table through an aluminium framed window. This is the only place in the house where it works reliably, even moving to the other side of the table causes it to have problems.
Well, with this fine weather, it must be time to get the spade out and start digging a ditch from the house to the garage and bury a couple of lengths of PVC or alkathene pipe and then run a couple of lengths of Cat5e cable through one of the pipes...
Lindsay Druett wrote:
Glenn Ramsey wrote:
I get 280KB/s using konqueror/fish on my 802.11b.
This link goes from my house to my garage/office and the link quality is not good because the garage has steel cladding. I have the access point taped to the window and I have line of sight to it from the kitchen table through an aluminium framed window. This is the only place in the house where it works reliably, even moving to the other side of the table causes it to have problems.
Well, with this fine weather, it must be time to get the spade out and start digging a ditch from the house to the garage and bury a couple of lengths of PVC or alkathene pipe and then run a couple of lengths of Cat5e cable through one of the pipes...
:-) it's on the todo list. Anyone have any gotchas about doing this? g -- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
Well, with this fine weather, it must be time to get the spade out and start digging a ditch from the house to the garage and bury a couple of lengths of PVC or alkathene pipe and then run a couple of lengths of Cat5e cable through one of the pipes...
:-) it's on the todo list. Anyone have any gotchas about doing this?
This is just from what I've read, and you probably know it anyway, but just in case... Try not to run parallel to power cables, or if you must, you'll have to look at shielded cable. According to the specs for Cat5e cable it'll run for 100m lengths at 100Mbit speeds, but it's different for Gigabit speeds, but generally if you think you might have more than 50m from computer to router/switch/hub/repeater/other computer and you want reliable Gigabit, you may want to opt for Cat6, otherwise Cat5e should be fine. -- Cameron
Camster342 wrote:
Well, with this fine weather, it must be time to get the spade out and start digging a ditch from the house to the garage and bury a couple of lengths of PVC or alkathene pipe and then run a couple of lengths of Cat5e cable through one of the pipes...
:-) it's on the todo list. Anyone have any gotchas about doing this?
This is just from what I've read, and you probably know it anyway, but just in case...
Try not to run parallel to power cables, or if you must, you'll have to look at shielded cable.
According to the specs for Cat5e cable it'll run for 100m lengths at 100Mbit speeds, but it's different for Gigabit speeds, but generally if you think you might have more than 50m from computer to router/switch/hub/repeater/other computer and you want reliable Gigabit, you may want to opt for Cat6, otherwise Cat5e should be fine.
I was originally going to bring up gotchas like plenty of sunscreen... Beer is an absolute requirement, really crucial if you intend to get other people to assist, and make sure the gas bottle is full and you have plenty of meat for the BBQ to feed those that assist. After the job, things like blisters, sore arms could be an outcome. But seriously, gig requirement is exactly the same as 10 Meg and 100 Meg and that is the cable run itself shouldn't exceed 90 Mtrs (allowing for 5mtrs each end for the patch cords), but the total distance is 100 Mtrs (Please note, metres, not feet). Cat5e vs. Cat6, for this sort of application, Cat5e is fine. Run in pipe/conduit, don't bury direct, also bury at ~ 600mm deep. When digging, be careful of existing stuff underground like power cables, gas/water pipes. Having running the cat5e/cat6 in it's own conduit would give sufficient segreation from power cables, but if you're running power in the same trench then go either side of the trench.
Hi Lindsay, thanks for all the tips. Lindsay Druett wrote:
Cat5e vs. Cat6, for this sort of application, Cat5e is fine. Run in pipe/conduit, don't bury direct, also bury at ~ 600mm deep.
Why does it need to be 600mm deep? I was planning on running the cable down a bit of 20mm alkathene pipe and burying it about 100mm deep. It will only be going across the lawn so unlikely to have vehicles, stock or otherwise going over it. Glenn -- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
Glenn Ramsey wrote:
Hi Lindsay, thanks for all the tips.
Lindsay Druett wrote:
Cat5e vs. Cat6, for this sort of application, Cat5e is fine. Run in pipe/conduit, don't bury direct, also bury at ~ 600mm deep.
Why does it need to be 600mm deep?
I was planning on running the cable down a bit of 20mm alkathene pipe and burying it about 100mm deep.
It will only be going across the lawn so unlikely to have vehicles, stock or otherwise going over it.
Glenn
600mm deep is regulation minimum depth (that includes data cable) unless it is going under concrete paths whereas then the requirement is for the cable to be deep enough so that if/when the path gets broken up the cable doesn't get damaged. 100mm is too shallow, 300mm would be the very least (considering you're going in a conduit), but why not 600mm ? The beer will taste better after the job is finished :)
Lindsay Druett wrote:
Glenn Ramsey wrote:
Hi Lindsay, thanks for all the tips.
Lindsay Druett wrote:
Cat5e vs. Cat6, for this sort of application, Cat5e is fine. Run in pipe/conduit, don't bury direct, also bury at ~ 600mm deep.
Why does it need to be 600mm deep?
I was planning on running the cable down a bit of 20mm alkathene pipe and burying it about 100mm deep.
It will only be going across the lawn so unlikely to have vehicles, stock or otherwise going over it.
Glenn
600mm deep is regulation minimum depth (that includes data cable) unless it is going under concrete paths whereas then the requirement is for the cable to be deep enough so that if/when the path gets broken up the cable doesn't get damaged.
100mm is too shallow, 300mm would be the very least (considering you're going in a conduit), but why not 600mm ? The beer will taste better after the job is finished :)
So there's no technical reason then? It's only for physical protection of the cable? Let's say I chose to not go with the regulations who is going smack my hand? I thought I was being generous with my 100 mm, there's plenty of other stuff that needs doing, besides digging an oversized but regulation trench, that will make the beer taste better. 50mm would be enough to keep it away from the lawnmower. Alkathene pipe is pretty tough, farmers often just lay on top of the ground and after a few months with grass growing over it and stock walking on it, it gets buried all by itself. I like that approach :-). Glenn -- Glenn Ramsey <glenn(a)componic.co.nz> 07 8627077 http://www.componic.co.nz
So there's no technical reason then? It's only for physical protection of the cable? Let's say I chose to not go with the regulations who is going smack my hand?
You could go with my approach; dig a spade depth around the edges and run unprotected Cat-5. Still going strong after nearly 5 years and one driveway replacement! http://bung.homeunix.net/gallery/driveway/Digger_029 Pulled the entire thing up and moved it when they came to dig the pit. Phone didn't survive too well though: http://bung.homeunix.net/gallery/driveway/Digger_025 Craig
On Sat, Aug 27, 2005 at 10:43:54AM +1200, DrWho? wrote:
Oliver Jones wrote:
I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little.
I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get.
That sounds about right.
No it doesn't; my laptop connects to a wireless AP which does NAT and then through another switch to my server (yes this is dumb but I can't be bothered getting it to not do NAT); iperf (as Kyle described elsewhere in the thread) says I get 20.3Mbits/sec between my laptop and my server. If I do a wget on a ~30megabyte file from the server to the laptop, I get 2.66MB/s, which is about 21Mbits/second. There is a wall or two between the laptop and the AP in a hallway closet about 10 metres away. It's possible that doing NAT and/or all those PCI busses on the router, switch and server are slowing it down a bit... John
I have set up quite a lot of wireless links. Generally I use Atheros chipset PCI cards in Gentoo Linux machines, and if I need an access point I use the Netgear WG302 as it is a solid piece of kit and very stable. On this network: Netgear WG302, 16Dbi Waveguide Duron 1.3, Gentoo Linux, Atheros 802.11G Card, Hills 19DBi Para Duron 1.3, Gentoo Linux, Atheros 802.11G Card, Hills 19DBi Para with about a kilometer (no more) between them I was getting 20mbit real throughput with iperf, and that was inside an OpenVPN tunnel. I also built a link using PC-Engines WRAP, Mikrotic RouterOS, 2x Atheros 802.11A Cards, 2x HyperLink 17DBi Backfire PC-Engines WRAP, Mikrotic RouterOS, 2x Atheros 802.11A Cards, 2x HyperLink 17DBi Backfire And got a full duplex 20mbit through it, over around 800 meters. That is with WPA turned on and in bridge mode. John R. McPherson wrote:
On Sat, Aug 27, 2005 at 10:43:54AM +1200, DrWho? wrote:
Oliver Jones wrote:
I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little.
I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get.
That sounds about right.
No it doesn't; my laptop connects to a wireless AP which does NAT and then through another switch to my server (yes this is dumb but I can't be bothered getting it to not do NAT); iperf (as Kyle described elsewhere in the thread) says I get 20.3Mbits/sec between my laptop and my server. If I do a wget on a ~30megabyte file from the server to the laptop, I get 2.66MB/s, which is about 21Mbits/second.
There is a wall or two between the laptop and the AP in a hallway closet about 10 metres away.
It's possible that doing NAT and/or all those PCI busses on the router, switch and server are slowing it down a bit...
John
------------------------------------------------------------------------
_______________________________________________ wlug mailing list | wlug(a)list.waikato.ac.nz Unsubscribe: http://list.waikato.ac.nz/mailman/listinfo/wlug
I just ran a quick iperf from my laptop to my nas storage server, theres a monowall box routing between the wireless subnet and wired subnet the server is on.. But this shouldn't affect the results as id expect the wireless network to be the slowest segment. This is a HP laptop very similar to yours (nx5000) in G mode to a linksyss WRT54gs, but I do have another 2 x wireless B devices on the network (and one is using remote desktop at the moment of this test being run) kyle(a)server:~$ iperf -c 10.3.68.2 ------------------------------------------------------------ Client connecting to 10.3.68.2, TCP port 5001 TCP window size: 16.0 KByte (default) ------------------------------------------------------------ [ 3] local 10.3.64.11 port 34591 connected with 10.3.68.2 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 19.3 MBytes 16.1 Mbits/sec kyle(a)server:~$ So you should definitely get some faster results than what you are seeing, I have noticed SCP's are always slow for me, so id recommend trying iperf and seeing what your results are. -----Original Message----- From: Oliver Jones [mailto:oliver(a)deeperdesign.com] Sent: Saturday, 27 August 2005 4:23 a.m. To: Waikato Linux Users Group Subject: [wlug] Wireless performance I've been twiddling with my wireless router settings and nothing I've changed so far has improved performance. Some appeared to degrade it a little. I'm still only getting 120kbytes/sec in gFTP/SSH. Which translates to about 1Mbit/sec. This is supposed to be a 54Mbit device. Come someone else with a 802.11g wireless network do a little throughput test and let me know what sort of data rate they get. Thanks. _______________________________________________ wlug mailing list | wlug(a)list.waikato.ac.nz Unsubscribe: http://list.waikato.ac.nz/mailman/listinfo/wlug
Thanks Kyle. This is the sort of info I need. Where can I get iperf? RPM for FC4 prefered. On Sat, 2005-08-27 at 22:14 +1200, Kyle Carter wrote:
I just ran a quick iperf from my laptop to my nas storage server, theres a monowall box routing between the wireless subnet and wired subnet the server is on.. But this shouldn't affect the results as id expect the wireless network to be the slowest segment.
You should put your B devices on different channels to your G devices if you can. Might require two AP's though.
This is a HP laptop very similar to yours (nx5000) in G mode to a linksyss WRT54gs, but I do have another 2 x wireless B devices on the network (and one is using remote desktop at the moment of this test being run)
kyle(a)server:~$ iperf -c 10.3.68.2 ------------------------------------------------------------ Client connecting to 10.3.68.2, TCP port 5001 TCP window size: 16.0 KByte (default) ------------------------------------------------------------ [ 3] local 10.3.64.11 port 34591 connected with 10.3.68.2 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 19.3 MBytes 16.1 Mbits/sec kyle(a)server:~$
So you should definitely get some faster results than what you are seeing, I have noticed SCP's are always slow for me, so id recommend trying iperf and seeing what your results are.
As an aside, I went to DSE today and got myself a few network cables which I've tried to discreetly run around the edge of my apartments lounge. I get around 8Mbytes to 9Mbytes a second between my Laptop and my NAS over the CAT6 cable when using gFTP (FTP protocol). The LinkSys WAG54G is operating as the switch. Regards On Sat, 2005-08-27 at 20:24 +1000, Oliver Jones wrote:
Thanks Kyle. This is the sort of info I need. Where can I get iperf? RPM for FC4 prefered.
On Sat, 2005-08-27 at 22:14 +1200, Kyle Carter wrote:
I just ran a quick iperf from my laptop to my nas storage server, theres a monowall box routing between the wireless subnet and wired subnet the server is on.. But this shouldn't affect the results as id expect the wireless network to be the slowest segment.
Try http://dast.nlanr.net/Projects/Iperf/ Ive moved away from fedora these days, but know ive had it running on fedora extremely simply. Ive moved to ubuntu just purely because of the great package sourcing. To get it installed on that box I just had to: apt-get install iperf I still find it so amazing ;-) Theres windows binaries too.. So you could try your laptop in windows to another machine also if it dual boots.. Might prove its not the wireless system and could be the driver support? -----Original Message----- From: Oliver Jones [mailto:oliver(a)deeperdesign.com] Sent: Saturday, 27 August 2005 10:24 p.m. To: Waikato Linux Users Group Subject: RE: [wlug] Wireless performance Thanks Kyle. This is the sort of info I need. Where can I get iperf? RPM for FC4 prefered. On Sat, 2005-08-27 at 22:14 +1200, Kyle Carter wrote:
I just ran a quick iperf from my laptop to my nas storage server, theres a monowall box routing between the wireless subnet and wired subnet the server is on.. But this shouldn't affect the results as id expect the wireless network to be the slowest segment.
You should put your B devices on different channels to your G devices if you can. Might require two AP's though.
This is a HP laptop very similar to yours (nx5000) in G mode to a linksyss WRT54gs, but I do have another 2 x wireless B devices on the network (and one is using remote desktop at the moment of this test being run)
kyle(a)server:~$ iperf -c 10.3.68.2 ------------------------------------------------------------ Client connecting to 10.3.68.2, TCP port 5001 TCP window size: 16.0 KByte (default) ------------------------------------------------------------ [ 3] local 10.3.64.11 port 34591 connected with 10.3.68.2 port 5001 [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 19.3 MBytes 16.1 Mbits/sec kyle(a)server:~$
So you should definitely get some faster results than what you are seeing, I have noticed SCP's are always slow for me, so id recommend trying iperf and seeing what your results are.
_______________________________________________ wlug mailing list | wlug(a)list.waikato.ac.nz Unsubscribe: http://list.waikato.ac.nz/mailman/listinfo/wlug
Oops.. Hit send before finishing
I just ran a quick iperf from my laptop to my nas storage server, theres a monowall box routing between the wireless subnet and wired subnet the server is on.. But this shouldn't affect the results as id expect the wireless network to be the slowest segment.
You should put your B devices on different channels to your G devices if you can. Might require two AP's though. Yeah, I only have 1 AP.. And don't really experience any problems.. It only seems to take a few minutes to empty my 1 gig CF card from my camera over the wireless and that's about the biggest datatransfer I do over the wireless. Anything bigger and I go plug it in.. (or just start it and walk away)
participants (12)
-
Andrew Thrift -
Camster342 -
Craig Box -
David Hallett -
DrWho? -
Glenn Ramsey -
Ian McDonald -
John R. McPherson -
Kyle Carter -
Lindsay Druett -
Matt Brown -
Oliver Jones