cancel
Showing results for 
Search instead for 
Did you mean: 

1Gbps Hub 4 - Huge ping spikes while downloading or uploading

bartkus05
On our wavelength

Hello,

So I put off writing this for a long while - but not its driving me more and more insane. So, when we're just gamiing, everything is fine - ping is between 20ms and 30ms which is quite fine. Problem begins when someone tries to upload or download something, even if its only downloading at 300Mbps or uploading at 30Mbps - the ping goes to 80ms - 100ms... Completely cripples the broadband. On a 1Gbps service this is ridicoulous, this should be able to download at 900Mbps and still not touch the pings.... I have a 2Gbps backbone in the house going to the main switch, which is then connected to a pfsense box and that goes to the hub 4. Pings on the local network up to the pfsense box are <1ms even with IPerf running at full speed - so no bottlenecks on my LAN.

Also, the RTT and RTTsd on the WAN are extremely high, 10ms / 3ms on average - lowest I've seen it go was 6ms / 1.8ms. I have tried 3 different NICs and the built in LAN port on the pfsense box and all produce the same results.

Is it the broadband that literally is so bad, or is it the hub 4? Perhaps its faulty? At this point, I'm literally willing to pay the penalty for leaving early and go back 80/20 from BT. You can't replace reliability and stability with just speed - which isn't 1Gbps at all either, most of the time it peaks out at 700Mbps.... 

And to confirm, the Hub 4 is running in Modem Mode - Router mode 1/2 my IoT WiFi devices refuse to connect to it lol

3.0 Downstream channels

Channel

Frequency (Hz)

Power (dBmV)

SNR (dB)

Modulation

Channel ID

25

331000000

0.7

39

QAM256

25

1

139000000

2.8

40.4

QAM256

1

2

147000000

2.7

40.4

QAM256

2

3

155000000

2.9

40.4

QAM256

3

4

163000000

2.6

40.9

QAM256

4

5

171000000

2.2

40.4

QAM256

5

6

179000000

2

40.4

QAM256

6

7

187000000

2.1

40.4

QAM256

7

8

195000000

2

40.4

QAM256

8

9

203000000

2

40.4

QAM256

9

10

211000000

1.9

40.4

QAM256

10

11

219000000

2

40.4

QAM256

11

12

227000000

1.9

40.4

QAM256

12

13

235000000

1.9

40.4

QAM256

13

14

243000000

2.1

40.9

QAM256

14

15

251000000

1.9

40.4

QAM256

15

16

259000000

2

40.4

QAM256

16

17

267000000

1.7

40.4

QAM256

17

18

275000000

1.5

40.4

QAM256

18

19

283000000

1.1

40.9

QAM256

19

20

291000000

1

40.4

QAM256

20

21

299000000

1.3

40.4

QAM256

21

22

307000000

1.1

40.4

QAM256

22

23

315000000

1

40.9

QAM256

23

24

323000000

0.7

39

QAM256

24

26

339000000

0.7

40.4

QAM256

26

27

347000000

0.5

39

QAM256

27

28

355000000

0.6

40.4

QAM256

28

29

363000000

0.6

40.4

QAM256

29

30

371000000

0.5

40.4

QAM256

30

31

379000000

0.5

40.4

QAM256

31

 

3.0 Downstream channels

Channel

Lock Status

RxMER (dB)

Pre RS Errors

Post RS Errors

25

Locked

38.983261

0

0

1

Locked

40.366287

0

0

2

Locked

40.366287

0

0

3

Locked

40.366287

0

0

4

Locked

40.946209

0

0

5

Locked

40.366287

0

0

6

Locked

40.366287

0

0

7

Locked

40.366287

0

0

8

Locked

40.366287

0

0

9

Locked

40.366287

0

0

10

Locked

40.366287

0

0

11

Locked

40.366287

0

0

12

Locked

40.366287

0

0

13

Locked

40.366287

0

0

14

Locked

40.946209

0

0

15

Locked

40.366287

0

0

16

Locked

40.366287

0

0

17

Locked

40.366287

0

0

18

Locked

40.366287

0

0

19

Locked

40.946209

0

0

20

Locked

40.366287

0

0

21

Locked

40.366287

0

0

22

Locked

40.366287

0

0

23

Locked

40.946209

0

0

24

Locked

38.983261

0

0

26

Locked

40.366287

0

0

27

Locked

38.983261

0

0

28

Locked

40.366287

0

0

29

Locked

40.366287

0

0

30

Locked

40.366287

0

0

31

Locked

40.366287

0

0

 

3.1 Downstream channels

Channel

Channel Width (MHz)

FFT Type

Number of Active Subcarriers

Modulation (Active Profile)

First Active Subcarrier (Hz)

33

96

4K

1880

QAM4096

759

 

3.1 Downstream channels

Channel ID

Lock Status

RxMER Data (dB)

PLC Power (dBmV)

Correcteds (Active Profile)

Uncorrectables (Active Profile)

33

Locked

42

1.9

111001074

0

 

3.0 Upstream channels

Channel

Frequency (Hz)

Power (dBmV)

Symbol Rate (ksps)

Modulation

Channel ID

1

39400000

36.5

5120 KSym/sec

64QAM

8

2

46200000

36.3

5120 KSym/sec

64QAM

7

3

53700000

36.3

5120 KSym/sec

64QAM

6

4

60300000

36.5

5120 KSym/sec

64QAM

5

 

3.0 Upstream channels

Channel

Channel Type

T1 Timeouts

T2 Timeouts

T3 Timeouts

T4 Timeouts

1

US_TYPE_STDMA

0

0

1

0

2

US_TYPE_STDMA

0

0

0

0

3

US_TYPE_STDMA

0

0

1

0

4

US_TYPE_STDMA

0

0

0

0

 

The average BMQ - I have no idea how this doesn't capture the ping spikes....

No Usage DayNo Usage Day

 

 

 

 

 

 

 

1 ACCEPTED SOLUTION

Accepted Solutions

Zak_M
Forum Team (Retired)
Forum Team (Retired)

Good afternoon @bartkus05

 

Thank you for coming back to us. 

 

I have managed to locate your account & I have managed to run some diagnostics from our side & from what I can see everything looks to be within the expected spec. 

 

Kind regards,

Zak_M

See where this Helpful Answer was posted

24 REPLIES 24

Andrew-G
Alessandro Volta

80-100ms loaded latency seems well within the realms of possibility.  Lower would be nice, but I can't see that a VDSL 80/20 connection would do better when somebody pulls down a Cleveland steamer.  What do a few fast.com tests report for loaded latency?  For comparison my connection reports 60-190ms, and usually in a range 90-120ms.

At the time you took the hub status data everything looks normal - power levels seem OK, SNR, modulation, error counts all where they'd be expected to be.  The BQM looks fairly normal for a DOCSIS connection - a few spikes, but not the sort of thing I'd expect to create routinely observable problems.  Do you see the same thing with the hub in router mode and your own router in access point mode?  I'm a tad puzzled by the 700 Mbps speeds you're seeing, perhaps the forum staff might spot something not visible to us, but they'll need the hub running in router mode to check the status.

I assume that you're concerned about the effect on gaming?  In which case if the forum staff can't see anything, maybe your only option is to run in a 35 Mbps Openreach connection purely for gaming?

 

Hey, thanks for your reply

I checked with fast.com, 850Mbps - 11ms unloaded 118ms loaded. I will check in router mode when I get a chance.

I have never seen anything like this on my vdsl connection - hence why the post. Even with someone downloading and saturating the bandwidth, ping never went above 20ms!! Only downside was 80Mbps speed.

I honestly thought it would be a faulty hub or another issue down the line. We had VM installed in March I believe and the cabinet is pure fibre, and fibre is up to the house then about 5m of coax. I was expecting much lower pings hehe but to see these spikes when download reaches 350mbps is ridiculous.

Another thing, pinging the gateway on the hub 4, so not the 192.168.100.1 but the 80...1, pings are in the 10ms average range. To 100.1 <1ms.

Hey,

Here's another fast.com result - directly connected to Hub 4 on LAN. Exactly the same as on Modem Mode (Maybe even slightly worse) 😞

Screenshot 2021-12-24 164248.png

It might be that the downstream QoS VM are doing in your area is HFQ so you max out the bandwidth the gateway can give you with others downloading in your area which will queue all packets to your WAN IP in one buffer.

but you should get 1Gb off peak times
---------------------------------------------------------------

bartkus05
On our wavelength

Hey, thanks for your reply. Sorry for a late come back to this - but I was tracking the speeds over a few days to put here. 

I very rarely see over 900Mbps - no matter the time of day. Its exactly the same in router mode. I have also upgraded all my networking gear past the Hub 4 to 2.5Gbps and CAT6A shielded cables (grounded properly). New cables between the Hub 4 and pfsense box as well. Absolutely no difference when it comes to the speed tests.

DateTimePing UnloadedPing LoadedDownload 
28/12/2021Morning11ms80ms800Mbps 
 Afternoon14ms111ms910Mbps
 Evening14ms23ms700Mbps
     
30/12/2021Morning21ms74ms310Mbps 
 Afternoon15ms250ms890Mbps
 Evening14ms56ms800Mbps
     
02/01/2022Morning9ms97ms1GbpsPages took ages to load this day - swapped to Google DNS from Cloudflare
 Afternoon21ms60ms910Mbps
 Evening20ms111ms910Mbps
     Upgraded to 2.5Gbps networking. All equipment replaced. Iperf running on pfsense shows full 2.5Gbps
05/01/2022Morning9ms122ms560Mbps
 Afternoon11ms119ms700Mbps
 Evening12ms141ms650Mbps

I didn't specify this earlier but I'm on the Gig1 + Home phone service, however the phone is not used at all, we just got this one to keep the number. Any chance someone from VM can run some tests on the line and see from their end?

What results do you get on this site?

https://www.waveform.com/tools/bufferbloat

 

Hey mate, here's the result. Ran it just now. 

I have also taken a read into bufferbloat - and out of curiosity set it up on my pfsense box using the combination of traffic limiters and download/upload queues. Exactly the same result on both runs - suggesting its not my pfsense box. Better results on this one compared to fast.com

bufferbloat test.png

I remain puzzled by the erratic download speeds (vast majority of the time VM deliver expected speeds very consistently), but in terms of latency and bufferbloat, what you're seeing is probably the limitations of the mix of analogue coax connections and the optical implementation.  A possible cause here is optical beat interference on the fibre side of things, that is a problem known to occur "in the wild" and affects the upstream and causes data corruption and packet loss, whether this is occurring for you, and whether VM have the expertise to detect and resolve it I couldn't say.   

OBI is a greater risk on a D3.1 upstream ( VM currently don't use D3.1 on the upstream), but lab tests have shown it occurring at low levels on D3.0 4 channel upstream setups, which is how VM setup their systems, and that might explain the erratic performance and latency issues that don't hit the BQM very obviously.  OBI is more likely on a high contention network, and VM do like to run as close as possible to available capacity, so it seems a credible guess. 

Perhaps somebody with real experience of VM's RFoG networks may be able to offer better informed comment.

So... I decided to go out and tighten all the coax connections around - including inside the brown box outside - made sure everything is plugged in properly. 

Coax connection was a little loose on the splitter where the power goes in to the Boostral unit - but not overly. 

I tried again replugging into different ports on the hub 4 - and the results I got were quite shocking tbh - Port 2/3 are dead in modem mode and router mode, I had nothing on them even after a restart, Port 1 works fine and now here's the results from Port 4 which I'm plugged into atm.

Possibly a faulty hub?

Screenshot 2022-01-07 153226.png