Yes, I have confidence in the routers themselves as everything from the routers testing back to anywhere else work as expected. It's the inside (NAT'd side of things that is giving us fits).
-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Scott Lambert Sent: Wednesday, July 09, 2014 6:39 PM To: Mikrotik discussions Subject: Re: [Mikrotik] Throughput Problem on Mikrotik RB951 I use a RB951Ui-2HnD wirelessly at my house. It is in my basement, connected via 100Mbps ethernet to the CCR1036 at the base of my tower which then crosses an AirFiber 24 link to the office. I can move as many bits per second as the wireless between my 4 year-old MacBook Pro and the 951 can handle. My downloads are usually between 10 and 30 M bits per second. It tends to depend on the site from which I am downloading. On Wed, Jul 09, 2014 at 03:47:31PM +0000, Justin Marshall wrote: > Having an issue with a few customers behind RB951 Mikrotiks. All of > these customers are on bridged Canopy 100's. > > Running a test from the Mikrotik (tools/bandwidth test) to a Mikrotik > we have setup at our office I'm seeing 8-10.2Mbps down. However when > the customer runs a speedtest (regardless of which speedtest server) > they usually max out around 3Mbps down. > I presume you run the MikroTik test, the customer runs the speedtest, you run the MikroTik test, ..., in rapid succession and the results follow the device all the time? Usually max out around 3Mbps? Do they *ever* max out above 3Mbps? What are the exact options you are using to test with from the mikrotik? Have you tested with TCP as well as UDP and gotten the same speeds? Put this on the PC and test with it: http://www.mikrotik.com/download/btest.exe It is not working under Wine on my Mac at the moment. I've not actually used it before. It seems to be less feature rich than the RouterOS built-in test. Winbox works like a treat under Wine. Using it should allow you to determine if the test methodology is making a difference in speed or if it is the device which is running the speed test, if it actually works. Also you can test between the PC and the on-site 951 as a control. Do large file downloads get the same rates as the speedtests? > One customer in particular was really helpful on trying to narrow down > what could be causing this. He tried 3 different computers (both > wired and wireless), 2 different cables between the computers and the > RB951, and different ports on the RB951. He also gave me teamviewer > access to one of these (wired to the RB951) to try a few different > things. > > > So far I tried: > > Upgraded to 6.15 firmware, did the /system routerboard upgrade. > > I've removed all traffic shaping rules that could possibly be > effecting their speed. > > Tried using a src-nat rule in lieu of masquerade. > > Also tried 2 different browsers to make sure it wasn't just a browser > issue on the computer I had teamviewer access to. > > Has anyone else experienced this, or have any other suggestions to > try? Where do you rate limit your customers, if you do? What do you use to rate limit your customers? If you hook a laptop directly behind the bridged Canopy 100, with the same IP as the 951 had, what speed does the laptop get? We don't have much Canopy gear; and I don't admin it. Ours is ancient, 900MHz, and only does 3Mbps down / 1Mbps up total. I presume yours is actually able to move the 10Mbps down. Are you tracking the throughput of your Canopy APs so that you know they have the headroom available while testing? Can you watch the, 1-3 second average, throughput on the SM and AP while the tests are running? And make sure there is no traffic to the tested SM before you run the tests? What does a /tool traceroute 64.250.34.225 look like? In RouterOS 6.recent you should be able to let it run for about 90 - 120 seconds to get a good sampling. Download WinMTR on the PC and run it for a similar period. Show us the results from both if nothing jumps out at you. Copy and paste the text, don't give us screenshots. I am not firing up a GUI mail client.... I'm a grumpy old fart. :-) The MTR results may indicate where, if anywhere, there is a bottleneck or packet loss. Re-run WinMTR and do a speedtest at the same time. We'll expect it to look worse. The extra stress on the links may make bottlenecks and packet loss more obvious. -- Scott Lambert KC5MLE Unix SysAdmin [email protected] _______________________________________________ Mikrotik mailing list [email protected] http://mail.butchevans.com/mailman/listinfo/mikrotik Visit http://blog.butchevans.com/ for tutorials related to Mikrotik RouterOS _______________________________________________ Mikrotik mailing list [email protected] http://mail.butchevans.com/mailman/listinfo/mikrotik Visit http://blog.butchevans.com/ for tutorials related to Mikrotik RouterOS
