What is the best way to test the reliability of an Internet connection?
I am getting lots of dropouts on things like Zoom meetings or Netflix, but a speed test is always healthy (typically close to 200Mbps on my Virgin connection). But speed isn't everything - a stop-start connection with lots of gaps can still get a high average speed.
Problem is I don't even know what to search for!
Ideally a simple "here's a rating" tool but happy to get my hands dirty if needed, eg if there are things I could monitor with Wireshark or whatever.
I have an important meeting tomorrow and at the moment I'm going to struggle to take part! -- Mark Rogers // More Solutions Ltd (Peterborough Office) // 0844 251 1450 Registered in England (0456 0902) 21 Drakes Mews, Milton Keynes, MK8 0ER
https://www.tecmint.com/linux-network-bandwidth-monitoring-tools/
On 08/02/2021 12:22, Mark Rogers wrote:
What is the best way to test the reliability of an Internet connection?
I am getting lots of dropouts on things like Zoom meetings or Netflix, but a speed test is always healthy (typically close to 200Mbps on my Virgin connection). But speed isn't everything - a stop-start connection with lots of gaps can still get a high average speed.
Problem is I don't even know what to search for!
Ideally a simple "here's a rating" tool but happy to get my hands dirty if needed, eg if there are things I could monitor with Wireshark or whatever.
I have an important meeting tomorrow and at the moment I'm going to struggle to take part! -- Mark Rogers // More Solutions Ltd (Peterborough Office) // 0844 251 1450 Registered in England (0456 0902) 21 Drakes Mews, Milton Keynes, MK8 0ER
main@lists.alug.org.uk http://www.alug.org.uk/ https://lists.alug.org.uk/mailman/listinfo/main Unsubscribe? See message headers or the web site above!
On Mon, 8 Feb 2021 at 12:26, Huge huge@huge.org.uk wrote:
https://www.tecmint.com/linux-network-bandwidth-monitoring-tools/
Thanks, I'll take a look
On Mon, Feb 08, 2021 at 12:22:49PM +0000, Mark Rogers wrote:
What is the best way to test the reliability of an Internet connection?
I am getting lots of dropouts on things like Zoom meetings or Netflix, but a speed test is always healthy (typically close to 200Mbps on my Virgin connection). But speed isn't everything - a stop-start connection with lots of gaps can still get a high average speed.
Problem is I don't even know what to search for!
Ideally a simple "here's a rating" tool but happy to get my hands dirty if needed, eg if there are things I could monitor with Wireshark or whatever.
Not a simple rating tool, but I'm a fan of smokeping (https://oss.oetiker.ch/smokeping/) for monitoring these things - clearly shows latency and packet loss over time (though nothing about speed) and you can set it up to monitor a few different end points.
J.
On Mon, 8 Feb 2021 at 12:54, Jonathan McDowell noodles@earth.li wrote:
Not a simple rating tool, but I'm a fan of smokeping (https://oss.oetiker.ch/smokeping/) for monitoring these things - clearly shows latency and packet loss over time (though nothing about speed) and you can set it up to monitor a few different end points.
Thanks, as it happens I have a PC on the network with Apache already installed so I've just apt install'ed SmokePing and I'll see what it tells me.
On Mon, 8 Feb 2021 12:22:49 +0000 Mark Rogers mark@more-solutions.co.uk allegedly wrote:
What is the best way to test the reliability of an Internet connection?
I am getting lots of dropouts on things like Zoom meetings or Netflix, but a speed test is always healthy (typically close to 200Mbps on my Virgin connection). But speed isn't everything - a stop-start connection with lots of gaps can still get a high average speed.
Problem is I don't even know what to search for!
Ideally a simple "here's a rating" tool but happy to get my hands dirty if needed, eg if there are things I could monitor with Wireshark or whatever.
I have an important meeting tomorrow and at the moment I'm going to struggle to take part! --
Mark
As Jonathan's reply suggests, you should probably be looking at latency problems. But smokeping drags in a lot of other dependencies (including apache) which you may not want.
Try the simple approach first and use ping, traceroute etc to see if there are any obvious slow spots in the network paths. A good alternative to straight ping is "oping" and its ncurses front end "noping". noping highlights any rtts it deems "aberrant".
If you are interestred in the latency of traffic to/from a webserver then try httping (which also has a nice ncurses front end.
Mick --------------------------------------------------------------------- Mick Morgan gpg fingerprint: FC23 3338 F664 5E66 876B 72C0 0A1F E60B 5BAD D312 https://baldric.net/about-trivia ---------------------------------------------------------------------
On Mon, 8 Feb 2021 at 14:03, mick mbm@rlogin.net wrote:
As Jonathan's reply suggests, you should probably be looking at latency problems. But smokeping drags in a lot of other dependencies (including apache) which you may not want.
I think latency might be the issue, but my problem is that most tools* like to give you a nice average rather than looking beyond that. A standard speed test would be helpful if as well as the average throughput it gave me min/max kbps figures but that seems so simple that there must be a reason why they don't. I tend to use speedtest-cli which is just a Python script so I might look into what it is actually doing.
* Most end user tools, anyway
Try the simple approach first and use ping, traceroute etc to see if there are any obvious slow spots in the network paths. A good alternative to straight ping is "oping" and its ncurses front end "noping". noping highlights any rtts it deems "aberrant".
noping is reporting times varying from 11ms to 43ms when run directly on the (OpenWRT on a Pi4) router. --- 8.8.8.8 ping statistics --- 133 packets transmitted, 133 received, 0.00% packet loss, time 2033.1ms RTT[ms]: min = 11, median = 14, p(95) = 23, max = 43
Is that good/bad?
If you are interestred in the latency of traffic to/from a webserver then try httping (which also has a nice ncurses front end.
--- http://www.google.com/ ping statistics --- 45 connects, 45 ok, 0.00% failed, time 46958ms round-trip min/avg/max = 37.8/49.5/65.8 ms
Again, with nothing to compare to I don't know how good/bad that is.
Thanks for the suggestions, I now feel I have more information and no more idea what the problem is :-)
On Mon, 8 Feb 2021 15:31:17 +0000 Mark Rogers mark@more-solutions.co.uk allegedly wrote:
noping is reporting times varying from 11ms to 43ms when run directly on the (OpenWRT on a Pi4) router. --- 8.8.8.8 ping statistics --- 133 packets transmitted, 133 received, 0.00% packet loss, time 2033.1ms RTT[ms]: min = 11, median = 14, p(95) = 23, max = 43
Is that good/bad?
That looks reasonable to me, though the max is a little on the high side. But it does depend where you do the test from - and remember that 8.8.8.8 is not a single fixed destination but is rather a distributed address used in large range of datacentres. Google's DNS servers are anycast systems placed so that they can be as close to as many end systems as possible. This means that /my/ ping to 8.8.8.8 may go to a completely different host to yours. (And because we use different intermediate routers our routes will be very different too).
Here's my result from my desktop (actually it goes over a VPN to a Heztner server in Germany)
--- 8.8.8.8 ping statistics --- 116 packets transmitted, 116 received, 0.00% packet loss, time 3749.2ms RTT[ms]: min = 32, median = 32, p(95) = 33, max = 34
and the same result directly from that VPN endpoint itself.
50 packets transmitted, 50 received, 0.00% packet loss, time 180.9ms RTT[ms]: min = 3, median = 4, p(95) = 4, max = 8
(so there is practically no latency there and the difference is caused by the additional multiple routers between my desktop and the endpoint).
If you are interestred in the latency of traffic to/from a webserver then try httping (which also has a nice ncurses front end.
--- http://www.google.com/ ping statistics --- 45 connects, 45 ok, 0.00% failed, time 46958ms round-trip min/avg/max = 37.8/49.5/65.8 ms
Again, with nothing to compare to I don't know how good/bad that is.
Again, goggle.com is not a single point. A better test would be to a webserver you know is /not/ on a CDN.
For example, my desktop resolves google.com to 216.58.200.14 which gives the following (pretty poor) results
--- http://google.com/ ping statistics --- 32 connects, 32 ok, 0.00% failed, time 51192ms round-trip min/avg/max = 612.7/630.7/907.1 ms
whereas my VM on Hetzner's network resolves google.com to 216.58.212.174 and gives the following (much better) result.
--- http://google.com/ ping statistics --- 50 connects, 50 ok, 0.00% failed, time 50207ms round-trip min/avg/max = 18.6/20.5/45.8 ms
a direct test to 216.58.212.174 from my desktop gives
--- http://216.58.212.174/ ping statistics --- 41 connects, 41 ok, 0.00% failed, time 43651ms round-trip min/avg/max = 75.0/76.8/87.2 ms
which is better than the first test to google.com from my desktop but not as good as the one from my VM.
And a test to google.de from my (Germany based VM) gives
--- http://google.de/ ping statistics --- 33 connects, 33 ok, 0.00% failed, time 33301ms round-trip min/avg/max = 18.5/20.5/29.7 ms
Thanks for the suggestions, I now feel I have more information and no more idea what the problem is :-)
The problem may lie in a number of possible places. You mentioned dropouts on zoom calls. Zoom uses a number of different ports (and both UDP and TCP as transports), see https://support.zoom.us/hc/en-us/articles/202342006-Network-Firewall-Setting... for example. If there are routers beween you and others on the call which prioritise zoom traffic differently you could get dropouts. It is also worth bearing in mind that ICMP is often given a much lower priority than other protocols by most devices so the ping tests you run will only give an /indication/ of possible problem spots.
Some ISPs are better than others too. TalkTalk had a pretty lousy reputation for network congestion at one time. Virginmedia is also known to have had problems in the past.
Ofcom reported on ISPs in 2017/2018 see https://www.ofcom.org.uk/research-and-data/telecoms-research/broadband-resea... which according to BroadbandGenie (https://www.broadbandgenie.co.uk/broadband/help/what-is-broadband-latency reported that EE, plusnet, BT and sky had latency of around 12ms whilst Virgin's was around 20ms - not a huge difference, but if that were coupled with additional problems elsewhere (possibly on your own network) it could become noticeable.
HTH
--------------------------------------------------------------------- Mick Morgan gpg fingerprint: FC23 3338 F664 5E66 876B 72C0 0A1F E60B 5BAD D312 https://baldric.net/about-trivia ---------------------------------------------------------------------
-
Huge -
Jonathan McDowell -
Mark Rogers -
mick