Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[phi2008]

I don't normally run the Thinkbroadband quality monitor so would appreciate some input. This is the "quality" graph I'm getting when I've shut down all the network traffic I can think of and left the computer alone -

And this is apparently what a good graph should look like -

At the moment I don't think the yellow latency streak is down to any activity on my router WAN interface. 

[les-70]

  I would say your graph is fairly normal.  I may be wrong but I think to to get the yellow to vanish the network needs to be totally free of the slightest congestion

[kitzuser87430]

What modem/router....the hg635 super hub?

I use that and if I run a ping from a command line every 30 or 40 seconds i receive a "rouge" high ping (about 60 -90 ms) I thhink the router priorities other tasks.

Code: [Select]

64 bytes from 188.65.113.193: icmp_seq=1 ttl=55 time=42.9 ms
64 bytes from 188.65.113.193: icmp_seq=2 ttl=55 time=83.2 ms
64 bytes from 188.65.113.193: icmp_seq=3 ttl=55 time=43.8 ms
64 bytes from 188.65.113.193: icmp_seq=4 ttl=55 time=43.7 ms
64 bytes from 188.65.113.193: icmp_seq=5 ttl=55 time=48.0 ms
64 bytes from 188.65.113.193: icmp_seq=6 ttl=55 time=45.0 ms
64 bytes from 188.65.113.193: icmp_seq=7 ttl=55 time=45.3 ms
64 bytes from 188.65.113.193: icmp_seq=8 ttl=55 time=43.3 ms
64 bytes from 188.65.113.193: icmp_seq=9 ttl=55 time=43.6 ms
64 bytes from 188.65.113.193: icmp_seq=10 ttl=55 time=44.5 ms
64 bytes from 188.65.113.193: icmp_seq=11 ttl=55 time=44.3 ms
64 bytes from 188.65.113.193: icmp_seq=12 ttl=55 time=43.7 ms
64 bytes from 188.65.113.193: icmp_seq=13 ttl=55 time=45.2 ms
64 bytes from 188.65.113.193: icmp_seq=14 ttl=55 time=43.7 ms
64 bytes from 188.65.113.193: icmp_seq=15 ttl=55 time=45.0 ms
64 bytes from 188.65.113.193: icmp_seq=16 ttl=55 time=43.6 ms
64 bytes from 188.65.113.193: icmp_seq=17 ttl=55 time=43.9 ms
64 bytes from 188.65.113.193: icmp_seq=18 ttl=55 time=48.7 ms
64 bytes from 188.65.113.193: icmp_seq=19 ttl=55 time=44.0 ms
64 bytes from 188.65.113.193: icmp_seq=20 ttl=55 time=44.5 ms
64 bytes from 188.65.113.193: icmp_seq=21 ttl=55 time=44.8 ms
64 bytes from 188.65.113.193: icmp_seq=22 ttl=55 time=44.1 ms
64 bytes from 188.65.113.193: icmp_seq=23 ttl=55 time=44.6 ms
64 bytes from 188.65.113.193: icmp_seq=24 ttl=55 time=44.3 ms
64 bytes from 188.65.113.193: icmp_seq=25 ttl=55 time=43.5 ms
64 bytes from 188.65.113.193: icmp_seq=26 ttl=55 time=45.5 ms
64 bytes from 188.65.113.193: icmp_seq=27 ttl=55 time=44.3 ms
64 bytes from 188.65.113.193: icmp_seq=28 ttl=55 time=43.6 ms
64 bytes from 188.65.113.193: icmp_seq=29 ttl=55 time=44.7 ms
64 bytes from 188.65.113.193: icmp_seq=30 ttl=55 time=43.4 ms
Before I switched to this router I received a normal ping graph.

Ian

[kitz]

this is apparently what a good graph should look like -

.. on a network with absolutely no traffic.

Not only that, a lot of routers have QoS with ICMP ping  assigned the low priority. In other words the router will always prioritise traffic such as http/ftp over icmp and concentrate on routing throughput traffic rather than responding to the 'are you there' 'yeah Im here' type of resquests.

The amount of yellow on your graph is perfectly acceptable on any network with QoS.   Red is the one to look out for, which is when packets have been dropped.  If you look closely, your connection has less droppped packets than the other so call perfect one

[phi2008]

Thanks for the replies ...

What modem/router....the hg635 super hub?

It's currently an HG612 modem plus a Ubiquiti EdgeRouter Lite, but earlier today it was a pfSense router with the HG635 Super Router set in bridge mode which gave pretty much the same result(actually the HG635 seemed somewhat better than the HG612 in connection stats). The only reason I've suddenly started wondering what is acceptable latency is because earlier today my usual pfSense router(dual core 1.8GHz Celeron) started giving very large latency responses and it turned out it there was some kind of software issue that was using 100% CPU - so since I couldn't track down the issue I "factory" reset it , checked the latency again and didn't expect to see much yellow on the graph ... 

Not only that, a lot of routers have QoS with ICMP ping  assigned the low priority. In other words the router will always prioritise traffic such as http/ftp over icmp and concentrate on routing throughput traffic rather than responding to the 'are you there' 'yeah Im here' type of resquests.

The graph is from a network with no traffic(or no significant Internet traffic I'm aware of).

[phi2008]

So here's my mtr output for the Thinkbroadband box -

Code: [Select]

                                                                                                Packets               Pings
 Host                                                                                            Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 192.168.1.10                                                                                  0.0%   400    0.3   0.2   0.1   0.7   0.0
 2. host-92-14-208-1.as43234.net                                                                  0.0%   400   13.7  13.6  13.2  16.3   0.2
 3. host-78-151-228-36.as13285.net                                                                0.0%   400   16.3  16.8  16.1  29.6   1.4
 4. host-78-151-225-192.static.as13285.net                                                        0.0%   399   16.8  18.6  16.2  79.9   7.8
 5. host-78-144-8-215.as13285.net                                                                 0.0%   399   17.8  19.0  16.8  49.8   5.4
 6. host-78-144-11-123.as13285.net                                                                0.0%   399   17.0  17.6  16.8  22.7   0.7
 7. talktalk-gw1.thdo.ncuk.net                                                                    0.0%   399   32.2  20.3  17.1 204.8  18.0
 8. te1-1-1-31.core-rs2.thdo.ncuk.net                                                             0.5%   399   17.7  20.4  17.3  72.6   5.1
 9. pingbox1.thinkbroadband.com                                                                   0.0%   399   23.1  17.7  17.0  23.4   0.8

talktalk-gw1.thdo.ncuk.net  seems the most notable by deviation, anyway it helps provide some explanation as to why I have a noticeable yellow topping on my nice green minimum. 

[Chrysalis]

We need to see the overnight data, in my opinion if its consistent 24/7 then thats due to ICMP QoS.  Which means its fine even tho doesnt look perfect on the graph.

If it decreases or dissapears at night it does suggest contention, however as long as you dont get packetloss and the blue bit remains steady it probably wont be performance affecting.

[boost]

So here's my mtr output for the Thinkbroadband box -

Code: [Select]

                                                                                                Packets               Pings
 Host                                                                                            Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 192.168.1.10                                                                                  0.0%   400    0.3   0.2   0.1   0.7   0.0
 2. host-92-14-208-1.as43234.net                                                                  0.0%   400   13.7  13.6  13.2  16.3   0.2
 3. host-78-151-228-36.as13285.net                                                                0.0%   400   16.3  16.8  16.1  29.6   1.4
 4. host-78-151-225-192.static.as13285.net                                                        0.0%   399   16.8  18.6  16.2  79.9   7.8
 5. host-78-144-8-215.as13285.net                                                                 0.0%   399   17.8  19.0  16.8  49.8   5.4
 6. host-78-144-11-123.as13285.net                                                                0.0%   399   17.0  17.6  16.8  22.7   0.7
 7. talktalk-gw1.thdo.ncuk.net                                                                    0.0%   399   32.2  20.3  17.1 204.8  18.0
 8. te1-1-1-31.core-rs2.thdo.ncuk.net                                                             0.5%   399   17.7  20.4  17.3  72.6   5.1
 9. pingbox1.thinkbroadband.com                                                                   0.0%   399   23.1  17.7  17.0  23.4   0.8

talktalk-gw1.thdo.ncuk.net  seems the most notable by deviation, anyway it helps provide some explanation as to why I have a noticeable yellow topping on my nice green minimum. 

You might be able to toggle UDP mode on a second instance and compare the two?

[phi2008]

We need to see the overnight data...

Here is a full day(ignore the activity before 6AM), with generally very little activity on my WAN interface -

So you can see latency picks up around 3PM, rises to 7-8PM, then decreases as the early morning approaches(this matches a partial previous graph I have).

EDIT:

Here is a graph from a Plusnet FTTC user for comparison -