Re: Throughput riddle
On Wed, 23 Mar 2016 18:48:23 -0700 David Christensenwrote: > On 03/21/2016 09:28 PM, Celejar wrote: > > On Sun, 20 Mar 2016 21:49:43 -0700 > > David Christensen wrote: > >> 4. The laptop TX bytes (1.2 GiB) and NAS RX bytes (731.5 MiB) do not > >> correlate well. > > > > Why would [laptop TX bytes and NAS RX bytes] match? > > You want to devise experiments that isolate and measure parameters for > which you have a hypothesis. My hypothesis was the laptop TX bytes > should match NAS RX bytes. > > > The data did not match my hypothesis, so the next step is to try to > figure out why. My point was that the laptop is transmitting lots of data to destinations other than the NAS, and the NAS has much less network activity going on, so why would we expect any correlation between those two figures? I think I understand what you mean. You're assuming a cold boot, so there would be no other traffic. These measurements have actually been taken from systems that have been up and running for a while. > >> 5. The NAS TX bytes (15.7 MiB) and laptop RX bytes (691.2 MiB) do not > >> correlate well. > > > > Again, why would they match? > > As above, but reversed -- NAS TX bytes and laptop RX bytes. > > > >> Please try another run per my assumptions, above. Revise and re-post > >> assumptions as necessary. Post new ifconfig, iwconfig, dmesg, and > >> anything else that looks like a clue. > > > > [Before testing:] > > > > Laptop: > > > > ifconfig > > > > RX packets:26691 errors:0 dropped:98 overruns:0 frame:0 > > Why were 98 laptop Rx packets were dropped before testing started? This is not from a cold boot, but from a running system. > >TX packets:16051 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:30347357 (28.9 MiB) TX bytes:2245418 (2.1 MiB) > > > > iwconfig > > > > Link Quality=67/70 Signal level=-43 dBm > >Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 > >Tx excessive retries:3577 Invalid misc:417 Missed beacon:0 > > What does laptop WiFi Tx excessive retries mean? Why were there 3577 Don't know. > before testing started? Again, this is not from a cold boot. > What does laptop WiFi Tx Invalid misc mean? Why were there 417 before > testing started? As above. > > Router (wlan0): > > > > RX packets:26691 errors:0 dropped:98 overruns:0 frame:0 > > Why were 98 router WiFi RX packets dropped before testing started? As above. > >TX packets:16051 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:30347357 (28.9 MiB) TX bytes:2245418 (2.1 MiB) > > > > Router (eth0): > > > > RX packets:1612778 errors:0 dropped:0 overruns:4594 frame:0 > > Why were 4594 router wired Rx packets overrun? As above. > >TX packets:1732474 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:439536079 (419.1 MiB) TX bytes:1175937292 (1.0 GiB) > >Interrupt:4 > > > > NAS: > > > > RX packets:552082 errors:0 dropped:36 overruns:0 frame:0 > > Why were 36 NAS RX packets dropped? As above. > >TX packets:252654 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:823241120 (785.1 MiB) TX bytes:18133067 (17.2 MiB) > >Interrupt:87 > > > > > > All of the above non-zero error statistics indicate problems that need > to be figured out. > > > > iperf: > > > > laptop -> NAS: 14.8/20.6/24.2 > > NAS -> laptop: 53.9/65.6/67.3 > > > > Hm, I never realized this - I'm getting 2-4 times the throughput when > > reversing the direction! > > > > laptop -> router: 8.74/10.2/11.4 > > router -> laptop: 53.4/56.2/57.1 > > > > Wow, 5-8 times the throughput when reversing! Thanks for the suggestion > > to reverse the tests - I really should have tried that on my own. > > > > Any idea what this means? > > The iperf numbers indicate that the router is having a hard time > receiving from the laptop. This could be due to a problem with the > laptop transmitter, the router receiver, or both. I have been coming to the tentative conclusion that the problem is with the laptop transmitter. Just now, I got: laptop -> router: 12.7/13.0/14.4 Mbps but from an android phone physically located right next to the laptop, I got: phone -> router: 35.4/40.3/43.9 ... Thanks for the suggestions, Celejar
Re: Throughput riddle
On 03/21/2016 09:28 PM, Celejar wrote: On Sun, 20 Mar 2016 21:49:43 -0700 David Christensenwrote: 4. The laptop TX bytes (1.2 GiB) and NAS RX bytes (731.5 MiB) do not correlate well. Why would [laptop TX bytes and NAS RX bytes] match? You want to devise experiments that isolate and measure parameters for which you have a hypothesis. My hypothesis was the laptop TX bytes should match NAS RX bytes. The data did not match my hypothesis, so the next step is to try to figure out why. 5. The NAS TX bytes (15.7 MiB) and laptop RX bytes (691.2 MiB) do not correlate well. Again, why would they match? As above, but reversed -- NAS TX bytes and laptop RX bytes. Please try another run per my assumptions, above. Revise and re-post assumptions as necessary. Post new ifconfig, iwconfig, dmesg, and anything else that looks like a clue. [Before testing:] Laptop: ifconfig RX packets:26691 errors:0 dropped:98 overruns:0 frame:0 Why were 98 laptop Rx packets were dropped before testing started? TX packets:16051 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:30347357 (28.9 MiB) TX bytes:2245418 (2.1 MiB) iwconfig Link Quality=67/70 Signal level=-43 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:3577 Invalid misc:417 Missed beacon:0 What does laptop WiFi Tx excessive retries mean? Why were there 3577 before testing started? What does laptop WiFi Tx Invalid misc mean? Why were there 417 before testing started? Router (wlan0): RX packets:26691 errors:0 dropped:98 overruns:0 frame:0 Why were 98 router WiFi RX packets dropped before testing started? TX packets:16051 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:30347357 (28.9 MiB) TX bytes:2245418 (2.1 MiB) Router (eth0): RX packets:1612778 errors:0 dropped:0 overruns:4594 frame:0 Why were 4594 router wired Rx packets overrun? TX packets:1732474 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:439536079 (419.1 MiB) TX bytes:1175937292 (1.0 GiB) Interrupt:4 NAS: RX packets:552082 errors:0 dropped:36 overruns:0 frame:0 Why were 36 NAS RX packets dropped? TX packets:252654 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:823241120 (785.1 MiB) TX bytes:18133067 (17.2 MiB) Interrupt:87 All of the above non-zero error statistics indicate problems that need to be figured out. iperf: laptop -> NAS: 14.8/20.6/24.2 NAS -> laptop: 53.9/65.6/67.3 Hm, I never realized this - I'm getting 2-4 times the throughput when reversing the direction! laptop -> router: 8.74/10.2/11.4 router -> laptop: 53.4/56.2/57.1 Wow, 5-8 times the throughput when reversing! Thanks for the suggestion to reverse the tests - I really should have tried that on my own. Any idea what this means? The iperf numbers indicate that the router is having a hard time receiving from the laptop. This could be due to a problem with the laptop transmitter, the router receiver, or both. Double-check whatever settings you have available for your WiFi interfaces in the laptop and in the router. Document them. [After testing:] ... Your statistics have information from at least four (4) iperf runs. You want to isolated each run, and use automation to get consistency/ repeatability -- e.g. use the scientific method. I'd suggest: 1. Write a script to gather data -- e.g. prints WiFi interface parameters, runs ifconfig, runs iwconfig, grep dmesg, etc.. 2. Write a script that sets up and runs one iperf test. 3. Write a script that calls #1 and saves the output to a file, calls #2 and saves the output to a file, calls #1 again and saves the output to a file, and then diff's the output of #1 before and #1 after. Then: 4. Cold boot everything. 5. Run script #3. 6. Repeat #4 and #5 until you get consistent results. 7. Change one thing and repeat #6, searching for correlation and/or causality. David
RE: Throughput riddle
Hi, [...] >>> It might be a good idea to upgrade to a dual-band access point, so >>> that you can use 5 GHz, which is typically has much cleaner channels. >> >> Thanks. See my other response in the thread regarding channel selection. >> >> Celejar >> > Also remember you can have too much RF power in your transmitter, as well > as your neighbours overloading your system, you could be overloading your own. > > Try turning down the transmitters power. That might indeed be a good idea because Another misconception that people do not take into account when increasing the power on the WiFi transmitter Although the transmitter can use a higher power setting and the clients can see the WiFi base tramsmitter, that does not mean the clients will use a higher setting. Therefore the base station might not hear the client. A lot of WiFi networks had to be "redesigned" when taking into account that a mobile phone uses a lot less power, usually aroud 60% of a laptop. Before the redesign a lot of times there would be failed communications a.k.a. noise when the WiFi base did not correctly hear the client, the client resending creating more noise, etc., degrading WiFi performance for all nearby devices. This can happen at home too. There is not really a difference in "Enterprise WiFi" and "SOHO WiFi" when you look at those parts. In just means that in Enterprise WiFi there is interference inside the corporate building, so it is the same company who is having the problems as the one that is causing them. With SOHO WiFi there is interference between you and your neighbours. Whether you care depends on whether you like them or if maybe because of your interference they are getting bigger / better WiFi transmitters as well and THEN bother you? ;-) It is best to adhere to "researched best practices" I think. Bonno Bloksma
Re: Throughput riddle
On Tue, 2016-03-22 at 00:45 -0400, Celejar wrote: > On Mon, 21 Mar 2016 12:15:56 - (UTC) > Dan Purgertwrote: > > > On 2016-03-18, Celejar wrote: > > > Hi, > > > > > > I'm trying to understand the throughput across the different links of > > > my little home network, and am perplexed by the measured wireless > > > throughput. > > > > > > The three main devices I'm interested in: > > > > > > Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). > > > Gigabit WAN and LAN, 802.11bgn wireless. > > > > I know I'm coming to the party a bit late, and it's probably been > > answered somewhere else in the posts here, but what channel width are > > you running here? > > > > If you're running 40 MHz, you WILL be getting combined Co-channel > > contention from all other devices on channels 1 and 6 (or 6 and 11). > > With 2.4 GHz connections, it's advisable to stick to 20 MHz channels, to > > limit the amount of contention you're getting. > > Running at 20 MHz. > > > > [...] > > > > > > Here's the part that baffles me - these are with the laptop connected > > > to the router wirelessly: > > > > > > Laptop - router: ~11.8 Mbps > > > > > > These numbers actually exhibit significant variance, but they're > > > generally at least this much, and at most about 15-20 Mbps. > > > > > > Laptop - NAS: ~14.7 Mbps > > > > > > Once again, these numbers vary widely, but are in line with the laptop > > > - router numbers. > > > > > > But here's the kicker: Ookla's speedtest (run on the laptop with > > > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > > > > > > [...] > > > > > > So the wireless link can apparently push at least 30 Mbps or so, so why > > > are my local wireless throughput numbers so much lower? > > > > Given that the router is a 2x2 device, have you tried sending multiple > > streams with iperf (with the "-P" client option)? As I (likely, > > incorrectly) recall, iperf defaults to one stream, whereas speedtest > > will run multiple streams. > > Just tried a few runs with "-P 2" - no difference. > > > > I was originally using one of the common 1/6/11 channels, and I switched > > > to 3 since I saw a lot of other stations on those channels. This may > > > have resulted in some improvement, but I'm still stuck locally as > > > above. What's the explanation for this - how can I possibly be getting > > > much better throughput to servers tens of miles away than to my local > > > stations? Does iperf somehow work fundamentally differently from > > > speedtest? If so, which is a better representation of actual throughput? > > > > Switch back to 1, 6, or 11 (and a 20 MHz channel). As others have > > explained, being on an "in-between" channel will result in you getting > > actual interference (rather than simply co-channel contention) ... not > > to mention causing interference for your neighbors. > > > > It might be a good idea to upgrade to a dual-band access point, so that > > you can use 5 GHz, which is typically has much cleaner channels. > > Thanks. See my other response in the thread regarding channel selection. > > Celejar > Also remember you can have too much RF power in your transmitter, as well as your neighbours overloading your system, you could be overloading your own. Try turning down the transmitters power. David.
Re: Throughput riddle
On Mon, 21 Mar 2016 12:15:56 - (UTC) Dan Purgertwrote: > On 2016-03-18, Celejar wrote: > > Hi, > > > > I'm trying to understand the throughput across the different links of > > my little home network, and am perplexed by the measured wireless > > throughput. > > > > The three main devices I'm interested in: > > > > Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). > > Gigabit WAN and LAN, 802.11bgn wireless. > > I know I'm coming to the party a bit late, and it's probably been > answered somewhere else in the posts here, but what channel width are > you running here? > > If you're running 40 MHz, you WILL be getting combined Co-channel > contention from all other devices on channels 1 and 6 (or 6 and 11). > With 2.4 GHz connections, it's advisable to stick to 20 MHz channels, to > limit the amount of contention you're getting. Running at 20 MHz. > > [...] > > > > Here's the part that baffles me - these are with the laptop connected > > to the router wirelessly: > > > > Laptop - router:~11.8 Mbps > > > > These numbers actually exhibit significant variance, but they're > > generally at least this much, and at most about 15-20 Mbps. > > > > Laptop - NAS: ~14.7 Mbps > > > > Once again, these numbers vary widely, but are in line with the laptop > > - router numbers. > > > > But here's the kicker: Ookla's speedtest (run on the laptop with > > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > > > > [...] > > > > So the wireless link can apparently push at least 30 Mbps or so, so why > > are my local wireless throughput numbers so much lower? > > Given that the router is a 2x2 device, have you tried sending multiple > streams with iperf (with the "-P" client option)? As I (likely, > incorrectly) recall, iperf defaults to one stream, whereas speedtest > will run multiple streams. Just tried a few runs with "-P 2" - no difference. > > I was originally using one of the common 1/6/11 channels, and I switched > > to 3 since I saw a lot of other stations on those channels. This may > > have resulted in some improvement, but I'm still stuck locally as > > above. What's the explanation for this - how can I possibly be getting > > much better throughput to servers tens of miles away than to my local > > stations? Does iperf somehow work fundamentally differently from > > speedtest? If so, which is a better representation of actual throughput? > > Switch back to 1, 6, or 11 (and a 20 MHz channel). As others have > explained, being on an "in-between" channel will result in you getting > actual interference (rather than simply co-channel contention) ... not > to mention causing interference for your neighbors. > > It might be a good idea to upgrade to a dual-band access point, so that > you can use 5 GHz, which is typically has much cleaner channels. Thanks. See my other response in the thread regarding channel selection. Celejar
Re: Throughput riddle
On Mon, 21 Mar 2016 11:41:05 + Bonno Bloksmawrote: > Hi, > > Responding to one part of your mail. The other parts have been covered in > other responses: > > > I was originally using one of the common 1/6/11 channels, and I switched to > > 3 since I saw a lot of other stations on those channels. > > This may have resulted in some improvement, but I'm still stuck locally as > > above. > > Be aware that by doing this you have created interference with even more wifi > networks then before. See https://en.wikipedia.org/wiki/List_of_WLAN_channels > If you select just channel 1, 6 or 11 you will have only interference with > other networks on that one channel. By selecting an "in between" channel 3 > you now have interference from both the channel 1 and the channel 6 networks. > Usually best is to see where the weakest networks are, on 1, 6 or 11 and > place your wifi there. > > In the 5GHz band there a lot more non overlapping channels and because of the > lower reach the chance of interference is lower as well. However, because of > that you might need more transmitters as well. I'm curious about this. I see that lots of sites claim this to be the correct approach to channel selection, apparently based largely on a Cisco articlef rom 2004: http://web.archive.org/web/20150502223736/http://www.cisco.com/c/en/us/td/docs/wireless/technology/channel/deployment/guide/Channel.html OTOH, some argue that these results are mainly relevant to enterprise contexts, but less so to typical consumer environments: http://superuser.com/questions/443178/is-it-better-to-use-a-crowded-2-4ghz-wi-fi-channel-1-6-11-or-unused-3-4-8 Everyone explains the theory as you've given it, and I suppose it makes sense, but theory is no substitute for actual empirical evidence. I suppose that for best results, I would have to benchmark throughput across my links while on different channels. And even if I get better throughput on 3, I'd still have to consider whether it's worth it, in light of possible negative effects to my neighbors. Thanks, > Bonno Bloksma Celejar
Re: Throughput riddle
On Sun, 20 Mar 2016 21:49:43 -0700 David Christensenwrote: > On 03/20/2016 07:10 PM, Celejar wrote: > > Laptop: > > > > RX packets:922215 errors:0 dropped:1967 overruns:0 frame:0 > >TX packets:1186319 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:724785210 (691.2 MiB) TX bytes:1311193642 (1.2 GiB) > > > > NAS: > > > > RX packets:509256 errors:0 dropped:3 overruns:0 frame:0 > >TX packets:234641 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:767110212 (731.5 MiB) TX bytes:16498728 (15.7 MiB) > >Interrupt:87 > > > > > > Router (wireless interface): > > > > RX packets:1474219 errors:0 dropped:0 overruns:0 frame:0 > >TX packets:1571154 errors:0 dropped:0 overruns:0 carrier:0 > >collisions:0 txqueuelen:1000 > >RX bytes:994970603 (948.8 MiB) TX bytes:1544655324 (1.4 GiB) > > Assuming: > > 1. The configuration is: > > laptop - WiFi - router - Cat 5 - NAS Yes, generally. [As per my original message, I occasionally connect the laptop to the router with ethernet cable for testing or other purposes.] > 2. You did a cold boot on everything. I did not. > 3. You ran iperf from the laptop to the NAS. Yes. > 4. You ran iperf from the NAS to the laptop. No. > Ideas: > > 1. I assume the dropped RX statistic means that received packets were > malformed, and not that they were dropped by a firewall rule (?). > > 2. The laptop WiFi interface dropping 1967 packets while the router > WiFi interface dropping zero leads me to think there's a problem that > only affects the laptop receiver. Perhaps you need turn up the > transmitter power in the router (?). > > 3. I don't understand why the NAS dropped 3 packets. A wired > connection should drop zero. Maybe it's the Cat 5 cable. Perhaps you > need a Cat 5E. > > 4. The laptop TX bytes (1.2 GiB) and NAS RX bytes (731.5 MiB) do not > correlate well. Why would they match? The laptop transmits lots of other stuff than iperf and backup traffic to the NAS. > 5. The NAS TX bytes (15.7 MiB) and laptop RX bytes (691.2 MiB) do not > correlate well. Again, why would they match? The laptop receives lots of stuff unrelated to the NAS. > 6. Information for the router wired interface is missing. > Please try another run per my assumptions, above. Revise and re-post > assumptions as necessary. Post new ifconfig, iwconfig, dmesg, and > anything else that looks like a clue. [Before testing:] Laptop: ifconfig RX packets:26691 errors:0 dropped:98 overruns:0 frame:0 TX packets:16051 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:30347357 (28.9 MiB) TX bytes:2245418 (2.1 MiB) iwconfig Link Quality=67/70 Signal level=-43 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:3577 Invalid misc:417 Missed beacon:0 Router (wlan0): RX packets:26691 errors:0 dropped:98 overruns:0 frame:0 TX packets:16051 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:30347357 (28.9 MiB) TX bytes:2245418 (2.1 MiB) Router (eth0): RX packets:1612778 errors:0 dropped:0 overruns:4594 frame:0 TX packets:1732474 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:439536079 (419.1 MiB) TX bytes:1175937292 (1.0 GiB) Interrupt:4 NAS: RX packets:552082 errors:0 dropped:36 overruns:0 frame:0 TX packets:252654 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:823241120 (785.1 MiB) TX bytes:18133067 (17.2 MiB) Interrupt:87 iperf: laptop -> NAS: 14.8/20.6/24.2 NAS -> laptop: 53.9/65.6/67.3 Hm, I never realized this - I'm getting 2-4 times the throughput when reversing the direction! laptop -> router: 8.74/10.2/11.4 router -> laptop: 53.4/56.2/57.1 Wow, 5-8 times the throughput when reversing! Thanks for the suggestion to reverse the tests - I really should have tried that on my own. Any idea what this means? [After testing:] laptop: ifconfig RX packets:434822 errors:0 dropped:115 overruns:0 frame:0 TX packets:291397 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen iwconfig Link Quality=60/70 Signal level=-50 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0 Tx excessive retries:233699 Invalid misc:1488 Missed beacon:0 router (wlan0): ifconfig RX packets:3983487 errors:0 dropped:0 overruns:0 frame:0 TX packets:5478517 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:1390191564 (1.2 GiB) TX bytes:2503989443 (2.3 GiB) router (eth0): RX packets:1861402 errors:0 dropped:0 overruns:4594
Re: Throughput riddle
On 2016-03-18, Celejarwrote: > Hi, > > I'm trying to understand the throughput across the different links of > my little home network, and am perplexed by the measured wireless > throughput. > > The three main devices I'm interested in: > > Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). > Gigabit WAN and LAN, 802.11bgn wireless. I know I'm coming to the party a bit late, and it's probably been answered somewhere else in the posts here, but what channel width are you running here? If you're running 40 MHz, you WILL be getting combined Co-channel contention from all other devices on channels 1 and 6 (or 6 and 11). With 2.4 GHz connections, it's advisable to stick to 20 MHz channels, to limit the amount of contention you're getting. > > [...] > > Here's the part that baffles me - these are with the laptop connected > to the router wirelessly: > > Laptop - router: ~11.8 Mbps > > These numbers actually exhibit significant variance, but they're > generally at least this much, and at most about 15-20 Mbps. > > Laptop - NAS: ~14.7 Mbps > > Once again, these numbers vary widely, but are in line with the laptop > - router numbers. > > But here's the kicker: Ookla's speedtest (run on the laptop with > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > > [...] > > So the wireless link can apparently push at least 30 Mbps or so, so why > are my local wireless throughput numbers so much lower? Given that the router is a 2x2 device, have you tried sending multiple streams with iperf (with the "-P" client option)? As I (likely, incorrectly) recall, iperf defaults to one stream, whereas speedtest will run multiple streams. > > I was originally using one of the common 1/6/11 channels, and I switched > to 3 since I saw a lot of other stations on those channels. This may > have resulted in some improvement, but I'm still stuck locally as > above. What's the explanation for this - how can I possibly be getting > much better throughput to servers tens of miles away than to my local > stations? Does iperf somehow work fundamentally differently from > speedtest? If so, which is a better representation of actual throughput? Switch back to 1, 6, or 11 (and a 20 MHz channel). As others have explained, being on an "in-between" channel will result in you getting actual interference (rather than simply co-channel contention) ... not to mention causing interference for your neighbors. It might be a good idea to upgrade to a dual-band access point, so that you can use 5 GHz, which is typically has much cleaner channels.
RE: Throughput riddle
Hi, Responding to one part of your mail. The other parts have been covered in other responses: > I was originally using one of the common 1/6/11 channels, and I switched to 3 > since I saw a lot of other stations on those channels. > This may have resulted in some improvement, but I'm still stuck locally as > above. Be aware that by doing this you have created interference with even more wifi networks then before. See https://en.wikipedia.org/wiki/List_of_WLAN_channels If you select just channel 1, 6 or 11 you will have only interference with other networks on that one channel. By selecting an "in between" channel 3 you now have interference from both the channel 1 and the channel 6 networks. Usually best is to see where the weakest networks are, on 1, 6 or 11 and place your wifi there. In the 5GHz band there a lot more non overlapping channels and because of the lower reach the chance of interference is lower as well. However, because of that you might need more transmitters as well. Bonno Bloksma
Re: Throughput riddle
On 03/20/2016 07:10 PM, Celejar wrote: Laptop: RX packets:922215 errors:0 dropped:1967 overruns:0 frame:0 TX packets:1186319 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:724785210 (691.2 MiB) TX bytes:1311193642 (1.2 GiB) NAS: RX packets:509256 errors:0 dropped:3 overruns:0 frame:0 TX packets:234641 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:767110212 (731.5 MiB) TX bytes:16498728 (15.7 MiB) Interrupt:87 Router (wireless interface): RX packets:1474219 errors:0 dropped:0 overruns:0 frame:0 TX packets:1571154 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:994970603 (948.8 MiB) TX bytes:1544655324 (1.4 GiB) Assuming: 1. The configuration is: laptop - WiFi - router - Cat 5 - NAS 2. You did a cold boot on everything. 3. You ran iperf from the laptop to the NAS. 4. You ran iperf from the NAS to the laptop. Ideas: 1. I assume the dropped RX statistic means that received packets were malformed, and not that they were dropped by a firewall rule (?). 2. The laptop WiFi interface dropping 1967 packets while the router WiFi interface dropping zero leads me to think there's a problem that only affects the laptop receiver. Perhaps you need turn up the transmitter power in the router (?). 3. I don't understand why the NAS dropped 3 packets. A wired connection should drop zero. Maybe it's the Cat 5 cable. Perhaps you need a Cat 5E. 4. The laptop TX bytes (1.2 GiB) and NAS RX bytes (731.5 MiB) do not correlate well. 5. The NAS TX bytes (15.7 MiB) and laptop RX bytes (691.2 MiB) do not correlate well. 6. Information for the router wired interface is missing. Please try another run per my assumptions, above. Revise and re-post assumptions as necessary. Post new ifconfig, iwconfig, dmesg, and anything else that looks like a clue. David
Re: Throughput riddle
On Sun, 20 Mar 2016 13:48:23 -0700 David Christensenwrote: > On 03/20/2016 11:07 AM, Celejar wrote: > > FWIW, I'm getting these: > > > > Tx excessive retries:392922 Invalid misc:5439 > > > > [Rx invalids are all 0] > > What machine? What file or tool? Laptop - reported by 'iwconfig' > What does 'ifconfig' report on the various machines? Laptop: RX packets:922215 errors:0 dropped:1967 overruns:0 frame:0 TX packets:1186319 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:724785210 (691.2 MiB) TX bytes:1311193642 (1.2 GiB) NAS: RX packets:509256 errors:0 dropped:3 overruns:0 frame:0 TX packets:234641 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:767110212 (731.5 MiB) TX bytes:16498728 (15.7 MiB) Interrupt:87 Router (wireless interface): RX packets:1474219 errors:0 dropped:0 overruns:0 frame:0 TX packets:1571154 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:994970603 (948.8 MiB) TX bytes:1544655324 (1.4 GiB) > David Celejar
Re: Throughput riddle
On 03/20/2016 11:07 AM, Celejar wrote: FWIW, I'm getting these: Tx excessive retries:392922 Invalid misc:5439 [Rx invalids are all 0] What machine? What file or tool? What does 'ifconfig' report on the various machines? David
Re: Throughput riddle
Hi On 21/03/2016 2:22 AM, arian wrote: > another simple bandwidth check: > # host1: nc -l 8090 > /dev/zero > # host2: dd if=/dev/zero | nc host1 8090 Okay, I'm trying the following. On host 1 # ssh -f name-in-config -L 18090:localhost:8090 'nc -l 8090 > /dev/zero' The "name-in-config" is setup in ~/.ssh/config file with host and port, direct login with the name works perfectly. The remote end [host 2] is running nc, I can see it via a different login. root 21335 21333 0 05:38 ? 00:00:00 bash -c nc -l 8090 > /dev/zero Back to host 1 # dd if=/dev/zero | nc localhost 18090 channel 3: open failed: connect failed: Connection refused Why do I get a connect failed on this? Kind Regards AndrewM signature.asc Description: OpenPGP digital signature
Re: Throughput riddle
On Sun, 20 Mar 2016 09:31:24 -0700 David Christensenwrote: ... > But, I still recommend Category 5E cables. > > > >> It's not clear if you are doing an apples-to-apples comparison. Perhaps > >> iperf isn't measuring what you think it is. > > > > That's exactly what I'm asking: what is iperf measuring, and why is it > > so much lower than the speedtest throughput? My understanding is that > > it simply measures straight-up TCP (or UPD, if desired) throughput. > > Even allowing for protocol overhead at the various network stack > > layers, the deviation shouldn't be that great. > > Perhaps you can find information on the project site (?): > > https://github.com/esnet/iperf ... > Assuming you've looked for error/ warning messages everywhere and > haven't seen anything obvious, the next step would seem to be enabling > or adding verbosity/ logging/ debugging/ etc., starting with iperf on > one end and ending with iperf on the other end. FWIW, I'm getting these: Tx excessive retries:392922 Invalid misc:5439 [Rx invalids are all 0] Which seems to mean that there are some problems with the connection. But I don't have any sense of what's normal, particularly with wireless, or how bad this is (obviously, I'd have to keep track of the retries / invalids per time / data transmitted). AFAICT, iperf has no verbosity / logging / debugging settings. Thanks, Celejar
Re: Throughput riddle
On Sun, 20 Mar 2016 16:22:58 +0100 arianwrote: > > > No - I've been using the default: 'iperf -c host' on laptop, 'iperf > > -sD' on router, NAS. > > > > Actually, this morning I've been getting about 17-20 Mbps between the > > laptop and NAS. I tried bidirectional testing ('iperf -d -c host', > > and the results actually remained constant. > > That sound weird. Can you check by means of tcpdump, wireshark or similar > wether packages are actually only hitting the air once? I'm not very skilled with wireshark, but I tried it. I see lots of packets with frame length 1514 (data length 1448) between the source and destination, interspersed with many 66 byte frame length packets between the dest and source (with no data). I suppose the former is the iperf payload data, and the latter is just TCP/IP acknowledgments. What should I be looking for, exactly? Okay. More tests - currently the iperf tests are showing 17-20 Mbps laptop - NAS, 14-16 Mbps laptop - router (even with '-d'). > another simple bandwidth check: > # host1: nc -l 8090 > /dev/zero > # host2: dd if=/dev/zero | nc host1 8090 2.0, 2.3, 2.4 and 2.5 Mbps laptop - NAS. Pretty much in line with the ones above. > bidirectional version: > # host1: nc -l 8090 > /dev/zero > # host2: nc -l 8090 | nc host1 8090 > # host1: dd if=/dev/zero | nc host2 8090 > > kill the last after a while > > with that I get very consistent 2.7 MB/s unidirectional, 1.4 MB/s > bidirectional I must be doing something wrong with the last - it just exists immediately, no error and no output. I think I understand the command syntax, and it makes sense to me, but it just doesn't work. I feel like I'm missing something obvious. Sorry if I'm being dense, thanks for the help, Celejar
Re: Throughput riddle
On 03/20/2016 06:10 AM, Celejar wrote: On Fri, 18 Mar 2016 18:01:58 -0700 David Christensenwrote: Perhaps the NAS has an automatic crossover feature on it's Gigabit port. If you do a computer-cable-computer test, you will want a (category 5E) crossover cable. If they weren't doing crossover, wouldn't I be doing a lot worse, or zero? Richard Hector clarified that crossover cables are only required for 10 and 100 Mbps. But, I still recommend Category 5E cables. It's not clear if you are doing an apples-to-apples comparison. Perhaps iperf isn't measuring what you think it is. That's exactly what I'm asking: what is iperf measuring, and why is it so much lower than the speedtest throughput? My understanding is that it simply measures straight-up TCP (or UPD, if desired) throughput. Even allowing for protocol overhead at the various network stack layers, the deviation shouldn't be that great. Perhaps you can find information on the project site (?): https://github.com/esnet/iperf None of your devices are running the firmware or OS the manufacturer intended for it. Perhaps you should revert the router and/or NAS, and test again. 1) The current situation is acceptable, and I'd rather run what I'm running at my current speeds than revert to the manufacturers' firmwares and OSs. 2) Intellectual curiosity is primarily what's driving me to understand my numbers. 3) If there's really something wrong with the OSS code, I should be filing bugs. Assuming you've looked for error/ warning messages everywhere and haven't seen anything obvious, the next step would seem to be enabling or adding verbosity/ logging/ debugging/ etc., starting with iperf on one end and ending with iperf on the other end. David
Re: Throughput riddle
> No - I've been using the default: 'iperf -c host' on laptop, 'iperf > -sD' on router, NAS. > > Actually, this morning I've been getting about 17-20 Mbps between the > laptop and NAS. I tried bidirectional testing ('iperf -d -c host', > and the results actually remained constant. That sound weird. Can you check by means of tcpdump, wireshark or similar wether packages are actually only hitting the air once? another simple bandwidth check: # host1: nc -l 8090 > /dev/zero # host2: dd if=/dev/zero | nc host1 8090 bidirectional version: # host1: nc -l 8090 > /dev/zero # host2: nc -l 8090 | nc host1 8090 # host1: dd if=/dev/zero | nc host2 8090 kill the last after a while with that I get very consistent 2.7 MB/s unidirectional, 1.4 MB/s bidirectional signature.asc Description: OpenPGP digital signature
Re: Throughput riddle
On Sun, 20 Mar 2016 14:43:00 +0100 arianwrote: > > > Here's the part that baffles me - these are with the laptop connected > > to the router wirelessly: > > > > Laptop - router:~11.8 Mbps > > Laptop - NAS: ~14.7 Mbps > > > > Once again, these numbers vary widely, but are in line with the laptop > > - router numbers. > > > > But here's the kicker: Ookla's speedtest (run on the laptop with > > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > > Just checking: you're not runnging iperf bidriectional, right? WiFi is half > duplex, in general for radio it's extremely difficult to achieve full duplex > when you don't have separate uplink/downlink carriers. > In that case you need to double your figures, and you'd get something > compatible with the speedtest results. No - I've been using the default: 'iperf -c host' on laptop, 'iperf -sD' on router, NAS. Actually, this morning I've been getting about 17-20 Mbps between the laptop and NAS. I tried bidirectional testing ('iperf -d -c host', and the results actually remained constant. I'm still getting only about 13 Mbps between the laptop and router. > regards, arian Celejar
Re: Throughput riddle
> Here's the part that baffles me - these are with the laptop connected > to the router wirelessly: > > Laptop - router: ~11.8 Mbps > Laptop - NAS: ~14.7 Mbps > > Once again, these numbers vary widely, but are in line with the laptop > - router numbers. > > But here's the kicker: Ookla's speedtest (run on the laptop with > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. Just checking: you're not runnging iperf bidriectional, right? WiFi is half duplex, in general for radio it's extremely difficult to achieve full duplex when you don't have separate uplink/downlink carriers. In that case you need to double your figures, and you'd get something compatible with the speedtest results. regards, arian signature.asc Description: OpenPGP digital signature
Re: Throughput riddle
On Sat, 19 Mar 2016 15:47:57 +1300 Richard Hectorwrote: ... > FWIW, most cabling professionals (of which definitely I'm not one) > don't make their own cables unless they absolutely have to. Factory > ones are so much more reliable. > > Riser cable, being intended for fixed installation, is solid core. The > appropriate cable for patch leads and other flexible applications is > stranded. To go with the 2 cable types, there are also different > connectors for each. Since most flexible cables are stranded, so are > most available connectors. If you're using solid cable with stranded > connectors, you're quite likely to get an unreliable connection. As > well as that, the solid cables are likely to fail sooner if they get > flexed more than they're designed for. > > More specifically, stranded connectors have spikes intended to go > through between the strands, while solid ones have springy things a > bit like the connectors in a mains socket (for flat blades), but much > smaller, that go either side and grip the solid wire. > > http://www.cableorganizer.com/articles/difference-between-solid-stranded-rj45-plugs.html Thanks for all this information. I'm simply using cables that came with my various devices, plus one or two that I actually bought. As per my earlier email, getting another few percent on iperf tests probably wouldn't translate into anything significant in my real world applications. > Richard Celejar
Re: Throughput riddle
On Fri, 18 Mar 2016 18:01:58 -0700 David Christensenwrote: > On 03/18/2016 09:48 AM, Celejar wrote: > > Hi, > > > > I'm trying to understand the throughput across the different links of > > my little home network, and am perplexed by the measured wireless > > throughput. > > > > The three main devices I'm interested in: > > > > Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). > > Gigabit WAN and LAN, 802.11bgn wireless. > > > > https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h > > > > Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn > > wireless. > > > > NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. > > > > https://archlinuxarm.org/platforms/armv5/seagate-goflex-net > > > > All throughput measurements taken with iperf (run three times and using > > the median result), unless specified otherwise. These first results are > > with the laptop connected to the router via cat5: > > > > Laptop - NAS: ~874 Mbps. > > I use category 5E cables for Gigabit. Category 5 and category 6 cables > were not reliable for me. Thanks - but as I indicated, 874 Mbps is close enough to the theoretical maximum that I don't think it would be worth it to me to start replacing cables. In real world usage, the difference for my applications would probably be minimal. > Perhaps the NAS has an automatic crossover feature on it's Gigabit port. > If you do a computer-cable-computer test, you will want a (category > 5E) crossover cable. If they weren't doing crossover, wouldn't I be doing a lot worse, or zero? ... > What happens if you connect two Gigabit computers through the router and > run iperf between the computers? Don't have another suitable computer that I can easily use to do this. > What happens if you connect one Gigabit computer and one 802.11n > computer through the router and run iperf between the computers? > > > What happens if you connect two 802.11n computer through the router and > run iperf between the computers? Ditto. > > Here's the part that baffles me - these are with the laptop connected > > to the router wirelessly: > > > > Laptop - router:~11.8 Mbps > > > > These numbers actually exhibit significant variance, but they're > > generally at least this much, and at most about 15-20 Mbps. > > > > Laptop - NAS: ~14.7 Mbps > > > > Once again, these numbers vary widely, but are in line with the laptop > > - router numbers. > > Troubleshooting WiFi is tough. Location and orientation of antennas is > critical. Understand that radio waves with a frequency of 2.4 GHz have > a wavelength of ~12 cm. One-quarter wavelength (~1 inch) can change > everything. You typically have little control over other WiFi devices > or other sources of RF interference. All matter in the universe absorbs > and re-radiates energy in curious ways, causing constructive/ > destructive interference patterns throughout space. So, Murphy's Law is > certain to strike. > > > Try repositioning the router. You want the antennas up high, radiating > horizontally (e.g. stick antennas oriented vertically), and far from all > other conductive materials -- metal-frame or masonry walls, electrical > wiring, metallic pipes, metallic ducts, HVAC grilles, metal-film > windows, metal window screens, televisions, other appliances, etc.. > Beware of items concealed inside wood-frame walls. Beware of metallic > drywall corner bead. > > > Test the laptop in several locations. Near-field radiation is more > complex than far-field, so use a wired connection if the laptop anywhere > near the router. A wired Gigabit connection is going to perform better > than 802.11n even under the best conditions, so consider running > category 5E cable to other usage locations. Adjust the router and > laptop radios for minimum power that gives reliable service. Thanks much for these suggestions. I know the basic issues (e.g., I have a DECT 6.0 cordless phone operating in the vicinity, although that apparently operates at 1.9 GHz and I don't know how much intereference that will cause with 2.4 GHz wifi), but I'm mainly puzzled by why the numbers are so much lower than the speedtest ones, as below. > > But here's the kicker: Ookla's speedtest (run on the laptop with > > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > > I'm paying Comcast for 25/5, and they apparently provision at > > 31.25/6.25, so I'm getting quite close to the theoretical max, even > > when the laptop is connected to the router wirelessly. Additionally, > > various Android phones also get close to the Comcast provisioned max > > when connecting wirelessly to the router. > > > > So the wireless link can apparently push at least 30 Mbps or so, so why > > are my local wireless throughput numbers so much lower? > > > > I was originally using one of the common 1/6/11 channels, and I switched > > to 3 since I saw a lot of other
Throughput riddle
Hi, I'm trying to understand the throughput across the different links of my little home network, and am perplexed by the measured wireless throughput. The three main devices I'm interested in: Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). Gigabit WAN and LAN, 802.11bgn wireless. https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn wireless. NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. https://archlinuxarm.org/platforms/armv5/seagate-goflex-net All throughput measurements taken with iperf (run three times and using the median result), unless specified otherwise. These first results are with the laptop connected to the router via cat5: Laptop - NAS: ~874 Mbps. I suppose this is close enough to the gigabit theoretical max, and there isn't any significant bottleneck. Router - NAS: ~217 Mbps Router - laptop:~198 Mbps Here the router CPU is apparently the bottleneck (top shows close to 100% CPU utilization by iperf for at least part of the 10 second iperf runs). I suppose that this is due to the bits needing to be copied out of the kernel networking stack into iperf's userspace memory, or something like that. I don't understanding why the NAS seems to be doing better, but I suppose it could be an artifact of the data. Here's the part that baffles me - these are with the laptop connected to the router wirelessly: Laptop - router:~11.8 Mbps These numbers actually exhibit significant variance, but they're generally at least this much, and at most about 15-20 Mbps. Laptop - NAS: ~14.7 Mbps Once again, these numbers vary widely, but are in line with the laptop - router numbers. But here's the kicker: Ookla's speedtest (run on the laptop with speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. I'm paying Comcast for 25/5, and they apparently provision at 31.25/6.25, so I'm getting quite close to the theoretical max, even when the laptop is connected to the router wirelessly. Additionally, various Android phones also get close to the Comcast provisioned max when connecting wirelessly to the router. So the wireless link can apparently push at least 30 Mbps or so, so why are my local wireless throughput numbers so much lower? I was originally using one of the common 1/6/11 channels, and I switched to 3 since I saw a lot of other stations on those channels. This may have resulted in some improvement, but I'm still stuck locally as above. What's the explanation for this - how can I possibly be getting much better throughput to servers tens of miles away than to my local stations? Does iperf somehow work fundamentally differently from speedtest? If so, which is a better representation of actual throughput? Celejar
Re: Throughput riddle
On Fri, 18 Mar 2016 12:44:30 -0600 (MDT) "John L. Ries" <jr...@salford-systems.com> wrote: > I don't know if it will help, but I hook up my Iomega NAS directly to my > desktop machine with a regular cat 5/6 cable (each has two gigabit > Ethernet ports, so each can connect to the rest of my network, as well > as to each other) and that seems to help the throughput by a lot (but I > don't have any numbers for you). So if your NAS has an extra Ethernet > port, you might want to hook it up to your laptop when you're in the > same room with it and use your wifi interface to connect to your > network. Certainly, you should avoid connecting to your NAS over wifi > if you're using it heavily, as that will definitely slow things down (it > seems that a lot more handshaking is required to connect through the air > than through a physical cable). Thanks. Currently, the NAS is used only as a backup target, so it's not a big deal - I'm mostly just frustrated and curious ... > On Friday 2016-03-18 10:48, Celejar wrote: > > >Date: Fri, 18 Mar 2016 10:48:24 > >From: Celejar <cele...@gmail.com> > >To: debian-user <debian-user@lists.debian.org> > >Subject: Throughput riddle > > > > Hi, > > > > I'm trying to understand the throughput across the different links of > > my little home network, and am perplexed by the measured wireless > > throughput. > > > > The three main devices I'm interested in: > > > > Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). > > Gigabit WAN and LAN, 802.11bgn wireless. > > > > https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h > > > > Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn > > wireless. > > > > NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. > > > > https://archlinuxarm.org/platforms/armv5/seagate-goflex-net > > > > All throughput measurements taken with iperf (run three times and using > > the median result), unless specified otherwise. These first results are > > with the laptop connected to the router via cat5: > > > > Laptop - NAS: ~874 Mbps. > > > > I suppose this is close enough to the gigabit theoretical max, and there > > isn't > > any significant bottleneck. > > > > Router - NAS: ~217 Mbps > > Router - laptop:~198 Mbps > > > > Here the router CPU is apparently the bottleneck (top shows close to > > 100% CPU utilization by iperf for at least part of the 10 second iperf > > runs). I suppose that this is due to the bits needing to be copied out > > of the kernel networking stack into iperf's userspace memory, or > > something like that. I don't understanding why the NAS seems to be > > doing better, but I suppose it could be an artifact of the data. > > > > Here's the part that baffles me - these are with the laptop connected > > to the router wirelessly: > > > > Laptop - router:~11.8 Mbps > > > > These numbers actually exhibit significant variance, but they're > > generally at least this much, and at most about 15-20 Mbps. > > > > Laptop - NAS: ~14.7 Mbps > > > > Once again, these numbers vary widely, but are in line with the laptop > > - router numbers. > > > > But here's the kicker: Ookla's speedtest (run on the laptop with > > speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > > I'm paying Comcast for 25/5, and they apparently provision at > > 31.25/6.25, so I'm getting quite close to the theoretical max, even > > when the laptop is connected to the router wirelessly. Additionally, > > various Android phones also get close to the Comcast provisioned max > > when connecting wirelessly to the router. > > > > So the wireless link can apparently push at least 30 Mbps or so, so why > > are my local wireless throughput numbers so much lower? > > > > I was originally using one of the common 1/6/11 channels, and I switched > > to 3 since I saw a lot of other stations on those channels. This may > > have resulted in some improvement, but I'm still stuck locally as > > above. What's the explanation for this - how can I possibly be getting > > much better throughput to servers tens of miles away than to my local > > stations? Does iperf somehow work fundamentally differently from > > speedtest? If so, which is a better representation of actual throughput? > > > > Celejar > > > > Celejar
Re: Throughput riddle
Forgot to mention: I use NetworkManager on the main box to configure the connection with the NAS (if you're using something else, consult your documentation). Under "IPv4 settings" (or IPv6 if you prefer), select "Link-local only" as the method. We expect that the NAS will do this automatically when it finds out the connection is point to point. --| John L. Ries | Salford Systems | Phone: (619)543-8880 x107 | or (435)867-8885 | --| On Friday 2016-03-18 12:44, John L. Ries wrote: Date: Fri, 18 Mar 2016 12:44:30 From: John L. Ries <jr...@salford-systems.com> To: Celejar <cele...@gmail.com> Cc: debian-user <debian-user@lists.debian.org> Subject: Re: Throughput riddle I don't know if it will help, but I hook up my Iomega NAS directly to my desktop machine with a regular cat 5/6 cable (each has two gigabit Ethernet ports, so each can connect to the rest of my network, as well as to each other) and that seems to help the throughput by a lot (but I don't have any numbers for you). So if your NAS has an extra Ethernet port, you might want to hook it up to your laptop when you're in the same room with it and use your wifi interface to connect to your network. Certainly, you should avoid connecting to your NAS over wifi if you're using it heavily, as that will definitely slow things down (it seems that a lot more handshaking is required to connect through the air than through a physical cable). --| John L. Ries | Salford Systems | Phone: (619)543-8880 x107 | or (435)867-8885 | --| On Friday 2016-03-18 10:48, Celejar wrote: Date: Fri, 18 Mar 2016 10:48:24 From: Celejar <cele...@gmail.com> To: debian-user <debian-user@lists.debian.org> Subject: Throughput riddle Hi, I'm trying to understand the throughput across the different links of my little home network, and am perplexed by the measured wireless throughput. The three main devices I'm interested in: Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). Gigabit WAN and LAN, 802.11bgn wireless. https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn wireless. NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. https://archlinuxarm.org/platforms/armv5/seagate-goflex-net All throughput measurements taken with iperf (run three times and using the median result), unless specified otherwise. These first results are with the laptop connected to the router via cat5: Laptop - NAS: ~874 Mbps. I suppose this is close enough to the gigabit theoretical max, and there isn't any significant bottleneck. Router - NAS: ~217 Mbps Router - laptop:~198 Mbps Here the router CPU is apparently the bottleneck (top shows close to 100% CPU utilization by iperf for at least part of the 10 second iperf runs). I suppose that this is due to the bits needing to be copied out of the kernel networking stack into iperf's userspace memory, or something like that. I don't understanding why the NAS seems to be doing better, but I suppose it could be an artifact of the data. Here's the part that baffles me - these are with the laptop connected to the router wirelessly: Laptop - router:~11.8 Mbps These numbers actually exhibit significant variance, but they're generally at least this much, and at most about 15-20 Mbps. Laptop - NAS: ~14.7 Mbps Once again, these numbers vary widely, but are in line with the laptop - router numbers. But here's the kicker: Ookla's speedtest (run on the laptop with speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. I'm paying Comcast for 25/5, and they apparently provision at 31.25/6.25, so I'm getting quite close to the theoretical max, even when the laptop is connected to the router wirelessly. Additionally, various Android phones also get close to the Comcast provisioned max when connecting wirelessly to the router. So the wireless link can apparently push at least 30 Mbps or so, so why are my local wireless throughput numbers so much lower? I was originally using one of the common 1/6/11 channels, and I switched to 3 since I saw a lot of other stations on those channels. This may have resulted in some improvement, but I'm still stuck locally as above. What's the explanation for this - how can I possibly be getting much better throughput to servers tens of miles away than to my local stations? Does iperf somehow work fundamentally differently from speedtest? If so, which is a better representation of actual throughput? Celejar
Re: Throughput riddle
On Fri, 18 Mar 2016 13:25:04 -0600 (MDT) "John L. Ries" <jr...@salford-systems.com> wrote: > Forgot to mention: > > I use NetworkManager on the main box to configure the connection with > the NAS (if you're using something else, consult your documentation). > Under "IPv4 settings" (or IPv6 if you prefer), select "Link-local only" > as the method. We expect that the NAS will do this automatically when > it finds out the connection is point to point. Thanks. > On Friday 2016-03-18 12:44, John L. Ries wrote: > > >Date: Fri, 18 Mar 2016 12:44:30 > >From: John L. Ries <jr...@salford-systems.com> > >To: Celejar <cele...@gmail.com> > >Cc: debian-user <debian-user@lists.debian.org> > >Subject: Re: Throughput riddle > > > > I don't know if it will help, but I hook up my Iomega NAS directly to my > > desktop machine with a regular cat 5/6 cable (each has two gigabit > > Ethernet ports, so each can connect to the rest of my network, as well > > as to each other) and that seems to help the throughput by a lot (but I > > don't have any numbers for you). So if your NAS has an extra Ethernet > > port, you might want to hook it up to your laptop when you're in the > > same room with it and use your wifi interface to connect to your > > network. Certainly, you should avoid connecting to your NAS over wifi > > if you're using it heavily, as that will definitely slow things down (it > > seems that a lot more handshaking is required to connect through the air > > than through a physical cable). > > > > --| > > John L. Ries | > > Salford Systems | > > Phone: (619)543-8880 x107 | > > or (435)867-8885 | > > ------| > > > > > > On Friday 2016-03-18 10:48, Celejar wrote: > > > >> Date: Fri, 18 Mar 2016 10:48:24 > >> From: Celejar <cele...@gmail.com> > >> To: debian-user <debian-user@lists.debian.org> > >> Subject: Throughput riddle > >> > >> Hi, > >> > >> I'm trying to understand the throughput across the different links of > >> my little home network, and am perplexed by the measured wireless > >> throughput. > >> > >> The three main devices I'm interested in: > >> > >> Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). > >> Gigabit WAN and LAN, 802.11bgn wireless. > >> > >> https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h > >> > >> Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn > >> wireless. > >> > >> NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. > >> > >> https://archlinuxarm.org/platforms/armv5/seagate-goflex-net > >> > >> All throughput measurements taken with iperf (run three times and using > >> the median result), unless specified otherwise. These first results are > >> with the laptop connected to the router via cat5: > >> > >> Laptop - NAS: ~874 Mbps. > >> > >> I suppose this is close enough to the gigabit theoretical max, and there > >> isn't > >> any significant bottleneck. > >> > >> Router - NAS: ~217 Mbps > >> Router - laptop: ~198 Mbps > >> > >> Here the router CPU is apparently the bottleneck (top shows close to > >> 100% CPU utilization by iperf for at least part of the 10 second iperf > >> runs). I suppose that this is due to the bits needing to be copied out > >> of the kernel networking stack into iperf's userspace memory, or > >> something like that. I don't understanding why the NAS seems to be > >> doing better, but I suppose it could be an artifact of the data. > >> > >> Here's the part that baffles me - these are with the laptop connected > >> to the router wirelessly: > >> > >> Laptop - router: ~11.8 Mbps > >> > >> These numbers actually exhibit significant variance, but they're > >> generally at least this much, and at most about 15-20 Mbps. > >> > >> Laptop - NAS: ~14.7 Mbps > >> > >> Once again, these numbers vary widely, but are in line with the laptop > >> - router numbers. > >> > >> But here's the kicker: Ookla's speedtest (run on the laptop with > >> speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. > >> I'm paying Comcast for 25/5, and they apparently provision at > >> 31.25/6.25, so I'm
Re: Throughput riddle
I don't know if it will help, but I hook up my Iomega NAS directly to my desktop machine with a regular cat 5/6 cable (each has two gigabit Ethernet ports, so each can connect to the rest of my network, as well as to each other) and that seems to help the throughput by a lot (but I don't have any numbers for you). So if your NAS has an extra Ethernet port, you might want to hook it up to your laptop when you're in the same room with it and use your wifi interface to connect to your network. Certainly, you should avoid connecting to your NAS over wifi if you're using it heavily, as that will definitely slow things down (it seems that a lot more handshaking is required to connect through the air than through a physical cable). --| John L. Ries | Salford Systems | Phone: (619)543-8880 x107 | or (435)867-8885 | --| On Friday 2016-03-18 10:48, Celejar wrote: Date: Fri, 18 Mar 2016 10:48:24 From: Celejar <cele...@gmail.com> To: debian-user <debian-user@lists.debian.org> Subject: Throughput riddle Hi, I'm trying to understand the throughput across the different links of my little home network, and am perplexed by the measured wireless throughput. The three main devices I'm interested in: Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). Gigabit WAN and LAN, 802.11bgn wireless. https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn wireless. NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. https://archlinuxarm.org/platforms/armv5/seagate-goflex-net All throughput measurements taken with iperf (run three times and using the median result), unless specified otherwise. These first results are with the laptop connected to the router via cat5: Laptop - NAS: ~874 Mbps. I suppose this is close enough to the gigabit theoretical max, and there isn't any significant bottleneck. Router - NAS: ~217 Mbps Router - laptop:~198 Mbps Here the router CPU is apparently the bottleneck (top shows close to 100% CPU utilization by iperf for at least part of the 10 second iperf runs). I suppose that this is due to the bits needing to be copied out of the kernel networking stack into iperf's userspace memory, or something like that. I don't understanding why the NAS seems to be doing better, but I suppose it could be an artifact of the data. Here's the part that baffles me - these are with the laptop connected to the router wirelessly: Laptop - router:~11.8 Mbps These numbers actually exhibit significant variance, but they're generally at least this much, and at most about 15-20 Mbps. Laptop - NAS: ~14.7 Mbps Once again, these numbers vary widely, but are in line with the laptop - router numbers. But here's the kicker: Ookla's speedtest (run on the laptop with speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. I'm paying Comcast for 25/5, and they apparently provision at 31.25/6.25, so I'm getting quite close to the theoretical max, even when the laptop is connected to the router wirelessly. Additionally, various Android phones also get close to the Comcast provisioned max when connecting wirelessly to the router. So the wireless link can apparently push at least 30 Mbps or so, so why are my local wireless throughput numbers so much lower? I was originally using one of the common 1/6/11 channels, and I switched to 3 since I saw a lot of other stations on those channels. This may have resulted in some improvement, but I'm still stuck locally as above. What's the explanation for this - how can I possibly be getting much better throughput to servers tens of miles away than to my local stations? Does iperf somehow work fundamentally differently from speedtest? If so, which is a better representation of actual throughput? Celejar
Re: Throughput riddle
On 19/03/16 14:01, David Christensen wrote: > I use category 5E cables for Gigabit. Category 5 and category 6 > cables were not reliable for me. Cat 5 cables _should_ work, in theory, though I gather some don't work so well. If you have any cat5 or better cables that are unreliable, I'd suspect the individual cable, not the stated spec. They may just be badly made. > Perhaps the NAS has an automatic crossover feature on it's Gigabit > port. If you do a computer-cable-computer test, you will want a > (category 5E) crossover cable. There's no need for crossover cables for gigabit. Gigabit communicates both ways over all 4 pairs anyway, and autonegotiating is part of the spec. > Make sure you are using the right cables and that they are known > good. I own an Ideal LinkMaster cable tester, and it has been worth > every penny: > > http://www.amazon.com/Linkmaster-UTP-stp-Cable-Tester/dp/B000LDC3LA A > tester like that will tell you if there's continuity in the right places - a cat 3 cable will test fine, and you could make a cable with phone cable, power cable or whatever you like and get it to test fine. Testing to Cat5 or whatever takes a _much_ more expensive tester, to check impedance and capacitance (and variations of those down the cable) and suchlike. It's a good start, but probably won't help much for "it's a bit slow". Richard
Re: Throughput riddle
On 03/18/2016 07:47 PM, Richard Hector wrote: FWIW, most cabling professionals (of which definitely I'm not one) don't make their own cables unless they absolutely have to. Factory ones are so much more reliable. Riser cable, being intended for fixed installation, is solid core. The appropriate cable for patch leads and other flexible applications is stranded. To go with the 2 cable types, there are also different connectors for each. Since most flexible cables are stranded, so are most available connectors. If you're using solid cable with stranded connectors, you're quite likely to get an unreliable connection. As well as that, the solid cables are likely to fail sooner if they get flexed more than they're designed for. More specifically, stranded connectors have spikes intended to go through between the strands, while solid ones have springy things a bit like the connectors in a mains socket (for flat blades), but much smaller, that go either side and grip the solid wire. http://www.cableorganizer.com/articles/difference-between-solid-stranded-rj45-plugs.html Thanks for the info! :-) David
Re: Throughput riddle
On 03/18/2016 06:47 PM, Richard Hector wrote: On 19/03/16 14:01, David Christensen wrote: I use category 5E cables for Gigabit. Category 5 and category 6 cables were not reliable for me. Cat 5 cables _should_ work, in theory, though I gather some don't work so well. If you have any cat5 or better cables that are unreliable, I'd suspect the individual cable, not the stated spec. They may just be badly made. Along with the tester, I also bought a 1,000 foot spool of category 5E riser cable, a crimping tool, and crimp connectors. Now I make my own cables and test them. :-) Perhaps the NAS has an automatic crossover feature on it's Gigabit port. If you do a computer-cable-computer test, you will want a (category 5E) crossover cable. There's no need for crossover cables for gigabit. Gigabit communicates both ways over all 4 pairs anyway, and autonegotiating is part of the spec. Thanks for the tip! :-) https://en.wikipedia.org/wiki/Ethernet_crossover_cable Make sure you are using the right cables and that they are known good. I own an Ideal LinkMaster cable tester, and it has been worth every penny: http://www.amazon.com/Linkmaster-UTP-stp-Cable-Tester/dp/B000LDC3LA A tester like that will tell you if there's continuity in the right places - a cat 3 cable will test fine, and you could make a cable with phone cable, power cable or whatever you like and get it to test fine. Testing to Cat5 or whatever takes a _much_ more expensive tester, to check impedance and capacitance (and variations of those down the cable) and suchlike. It's a good start, but probably won't help much for "it's a bit slow". I've spoken to sound/ communications technicians who talked about validating Ethernet cabling to X MHz and I've glanced at the expensive Fluke networking meters that I assume can do such measurements. Fortunately, a simple continuity tester and whatever utilities the OS provides have been enough to get me by. David
Re: Throughput riddle
On 03/18/2016 09:48 AM, Celejar wrote: Hi, I'm trying to understand the throughput across the different links of my little home network, and am perplexed by the measured wireless throughput. The three main devices I'm interested in: Router: Buffalo WZR-HP-G300NH running OpenWrt (Chaos Calmer 15.05). Gigabit WAN and LAN, 802.11bgn wireless. https://wiki.openwrt.org/toh/buffalo/wzr-hp-g300h Laptop: Thinkpad T61 running Jessie 8.3. Gigabit ethernet, 802.11abgn wireless. NAS: Seagate GoFlex Net [STAK100] runninng Debian Jessie 8.3. https://archlinuxarm.org/platforms/armv5/seagate-goflex-net All throughput measurements taken with iperf (run three times and using the median result), unless specified otherwise. These first results are with the laptop connected to the router via cat5: Laptop - NAS: ~874 Mbps. I use category 5E cables for Gigabit. Category 5 and category 6 cables were not reliable for me. Perhaps the NAS has an automatic crossover feature on it's Gigabit port. If you do a computer-cable-computer test, you will want a (category 5E) crossover cable. Make sure you are using the right cables and that they are known good. I own an Ideal LinkMaster cable tester, and it has been worth every penny: http://www.amazon.com/Linkmaster-UTP-stp-Cable-Tester/dp/B000LDC3LA I suppose this is close enough to the gigabit theoretical max, and there isn't any significant bottleneck. Router - NAS: ~217 Mbps Router - laptop:~198 Mbps Here the router CPU is apparently the bottleneck (top shows close to 100% CPU utilization by iperf for at least part of the 10 second iperf runs). I suppose that this is due to the bits needing to be copied out of the kernel networking stack into iperf's userspace memory, or something like that. I don't understanding why the NAS seems to be doing better, but I suppose it could be an artifact of the data. The numbers are within 10%, so I'd call them the same as far as router performance is concerned. What happens if you connect two Gigabit computers through the router and run iperf between the computers? What happens if you connect one Gigabit computer and one 802.11n computer through the router and run iperf between the computers? What happens if you connect two 802.11n computer through the router and run iperf between the computers? Here's the part that baffles me - these are with the laptop connected to the router wirelessly: Laptop - router:~11.8 Mbps These numbers actually exhibit significant variance, but they're generally at least this much, and at most about 15-20 Mbps. Laptop - NAS: ~14.7 Mbps Once again, these numbers vary widely, but are in line with the laptop - router numbers. Troubleshooting WiFi is tough. Location and orientation of antennas is critical. Understand that radio waves with a frequency of 2.4 GHz have a wavelength of ~12 cm. One-quarter wavelength (~1 inch) can change everything. You typically have little control over other WiFi devices or other sources of RF interference. All matter in the universe absorbs and re-radiates energy in curious ways, causing constructive/ destructive interference patterns throughout space. So, Murphy's Law is certain to strike. Try repositioning the router. You want the antennas up high, radiating horizontally (e.g. stick antennas oriented vertically), and far from all other conductive materials -- metal-frame or masonry walls, electrical wiring, metallic pipes, metallic ducts, HVAC grilles, metal-film windows, metal window screens, televisions, other appliances, etc.. Beware of items concealed inside wood-frame walls. Beware of metallic drywall corner bead. Test the laptop in several locations. Near-field radiation is more complex than far-field, so use a wired connection if the laptop anywhere near the router. A wired Gigabit connection is going to perform better than 802.11n even under the best conditions, so consider running category 5E cable to other usage locations. Adjust the router and laptop radios for minimum power that gives reliable service. But here's the kicker: Ookla's speedtest (run on the laptop with speedtest-cli) shows 29.01/5.89 (d/u), and this is fairly consistent. I'm paying Comcast for 25/5, and they apparently provision at 31.25/6.25, so I'm getting quite close to the theoretical max, even when the laptop is connected to the router wirelessly. Additionally, various Android phones also get close to the Comcast provisioned max when connecting wirelessly to the router. So the wireless link can apparently push at least 30 Mbps or so, so why are my local wireless throughput numbers so much lower? I was originally using one of the common 1/6/11 channels, and I switched to 3 since I saw a lot of other stations on those channels. This may have resulted in some improvement, but I'm still stuck locally as above. What's the explanation for this - how can I possibly be getting
Re: Throughput riddle
On 19/03/16 15:07, David Christensen wrote: > On 03/18/2016 06:47 PM, Richard Hector wrote: >> On 19/03/16 14:01, David Christensen wrote: >>> I use category 5E cables for Gigabit. Category 5 and category >>> 6 cables were not reliable for me. >> >> Cat 5 cables _should_ work, in theory, though I gather some don't >> work so well. If you have any cat5 or better cables that are >> unreliable, I'd suspect the individual cable, not the stated >> spec. They may just be badly made. > > Along with the tester, I also bought a 1,000 foot spool of category > 5E riser cable, a crimping tool, and crimp connectors. Now I make > my own cables and test them. :-) FWIW, most cabling professionals (of which definitely I'm not one) don't make their own cables unless they absolutely have to. Factory ones are so much more reliable. Riser cable, being intended for fixed installation, is solid core. The appropriate cable for patch leads and other flexible applications is stranded. To go with the 2 cable types, there are also different connectors for each. Since most flexible cables are stranded, so are most available connectors. If you're using solid cable with stranded connectors, you're quite likely to get an unreliable connection. As well as that, the solid cables are likely to fail sooner if they get flexed more than they're designed for. More specifically, stranded connectors have spikes intended to go through between the strands, while solid ones have springy things a bit like the connectors in a mains socket (for flat blades), but much smaller, that go either side and grip the solid wire. http://www.cableorganizer.com/articles/difference-between-solid-stranded-rj45-plugs.html Richard