Speed drops and streaming unreliability, regularly to near zero

[#1]

I have had rolling problems for the last 3 years, that were not present when I was on a BT contract. I have a seemingly OK sync speed (BT gave me 60+/12+, Plusnet give me 39-45/12+) - which is fine. No line drops were recoded but things like VPN links and streaming links would drop out 20+ times a day. Over the pandemic that became a nightmare. The problem continues.

I have previously had the Openreach folk out to check my house wiring and they've confirmed the internal wiring isn't causing a fault. They also switched pairs to the cabinet to make sure it wasn't the last bit of the chain from the cabinet, but that still goes  through a massive dropwire to a pole, which swings wildly in the wind (and the fault occurs more in the wind/rain, but that could be placebo).

Well I recently installed some software running on a hard-wired machine, which runs a speed test every hour. Here's an indication of my problem:

The last week's results are shown in the attachment. The "low" values are fairly frequent, and that's just sampling once per hour. When speaking to openreach or customer services, it's always "well it's testing fine now and there's no sign of a line drop" but when services are crippled by sudden drops in connection quality, the effect is the same.

[#2]

@Coffeeking 

I find it disappointing that a nearly a month has passed and there has been no staff response here.

The monitoring is interesting but shows nothing helpful beyond the measurable speed is variable.  There is no discernible time base pattern ... unless you can correlate the bad spells with bad weather.

Which router are you using?  If you are using the Plusnet Hub One, then that is the same as the BYHH5a, for which there is a custom version of router stats - see the link below.  I would recommend setting that up and running it 24x7 and then lets review the SNRM plots.  I suggest a 10 second sample interval with 720 points per graph.

Any questions, please ask.

If the router you are using permits response to external pings, setting up a Think Broadband Quality Monitor can be useful ... but if you are on a dynamic IP address, you need to watch out for drops an a new IP address being assigned.  Yes that can be a bit of a pain, but maybe less pain than living with the problem you are experiencing?

[#3]

Hey @Coffeeking,

I'm really sorry that we haven't been able to get back in touch with you sooner, we've been busier than anticipated recently and it appears the OP didn't get picked up. I've had a look over the account and I noticed that the router's been online without interference for the last 100 days. I also noticed that the last record we have of being contacted about speed issues was from September of 2021 - would you say this problem has existed for the entire interim period, or does it only seem to be an issue sporadically?

If it's the latter, don't worry, we can still help - it's just good to know as much as possible before proceeding with an investigation. Speaking of which, before we do go any further, please could you try turning the router off completely (at the wall too, if you can) for about 2 minutes and then back on again?

[#4]

Alex,

That’s a really helpful update - 100 days without a PPP drop is amazing - that points to a very stable line. Is it possible that this user is associated with a hot back haul at the exchange?

The other considerations become other concurrent activity during a speed test or WiFi performance challenges.

[#5]

I've logged a request for a DLM Reset, as it looked like a cap was in place, but other than this everything seems to be performing as we'd expect. Obviously, with the router being up and running for 100 days, the line seems stable, so hopefully once the resets completed the speed naturally improves. If it doesn't I'd still recommend a reboot, so we can see if it has any affect.

[#6]

I'm still seeing problems with this I'm afraid, nothing ever changes. 

Every so often I lose either upload, download or both speeds for sometimes 10 minutes, sometimes several hours at a time.

I called late last month and was told the line checked out as faulty and an engineer would be sent out to fix it without needing access to the property.  They did change something, I wasn't at the property, but then they sent me an email saying I should run a speed check to test it, then 6 minutes later sent me another saying because I didn't run a speed check I wasn't able to claim they were not meeting performance guarantee. I was at work, in another country. I could have done the check if they'd given more than a few minutes.

But still, today, 3 hours so far with next to zero download speed (see attachments, one showing download speed over the month with the very obvious fault and repair) and then the ongoing nonsense. It's been like this for 3 years now, I've lost track of the number of times I've had this conversation and gone through the "is it in the master socket.....is it checked via hardwired connection". This only started happening when I switched to Plusnet, from BT, which I thought was a safe bet. Every time I'm told there's no problem and the modem isn't disconnecting, so there should be no problems. But I cannot use my work VPN from here because it drops out every 15 minutes due to some unknown fault. If I use my VPN via my neighbours BT connection it runs all day without failing. If I use my partner's connection in another city, it runs all day, so the fault seems to be with the transport of data through a seemingly "stable" plusnet connection, which just gets throttled to death by some unknown fault. 

There's only so many hours in a day, and complaining and trying to get a fix for an intermittent fault where you're told every time that you're imagining it as there's no fault, or it must be wifi interference (despite being connected by cat6), means I don't complain at every occurrence. If I did, I'd have had a heart attack by now, so I only complain when it's bad and I have spare time. 

To round up my view of it, which I realise is maybe not as complete as the tech's view from that end:

1) Stable "connection" - modem doesn't drop connection very often, if at all.

2) Data flow grinds to a halt - sometimes on upload, sometimes on download. 

3) Measured on wired connection, on the master socket and not, using a draytek separate modem and a D-Link router (because the Pnet router wouldn't let me configure the security I needed, and I hoped sending 100 quid on a modem might improve the reliability).

4) Disconnecting and re-connecting can fix it, but doesn't always.

5) Problem often (but not always) gets worse with windy or wet/windy weather (originally I questioned the dropwire which has a huge span but it was clear of all the trees when this started and doesn't move a lot in the wind). But it still is a problem in the dry sunny weather.

6) Previously an engineer has checked out the internal wiring and gave it the thumbs up before he swapped me to a different pair in the cabinet. They asked me to check with the modem on the master socket, and not, and found no difference. 

7) Every time, I'm told the problem is fixed, and it's not.

I spend a lot of time away with work, so I can't always get access to the hardware to run tests, though I can usually dial in remotely to a hard-wired linux machine. While I'm away, I use my home connection to connect to my personal network hardware for y hobby use, and use the items in the network, when I can't access it due to slow up or download speeds, I'm basically stuffed.

[#7]

Something is not ringing true here.

Some key questions asked have not been answered, such as which router is being used?

Historic 100+ days PPP session point to a very stable line.

The basics need to be revisited here...

Is the telephone line quite - dial 17070 option 2.  If you do not have a handset, please borrow one.  Your ear on the line is the best diagnostic tool for ruling out (especially intermittent) noise on the line.

Can you please post the full router stats, including attenuation, SNRM and if available error stats.

If a Plusnet router, the firmware level.

What else is in your network - WiFi APs, switches, ethernet over power adapters.

How are the devices encountering issues connecting to the internet?  Ethernet connected of wifi connected?

If WiFi which band and if there is more than one AP, can you be sure which AP they connect to (that of the router or one of the others)?

@James30 - could you look back to see if Alex looked into the possibility of a hot VLAN please?

[#8]

Indeed I missed the original response, I'm busy so sometimes they slip by.

DrayTek Vigor 130 is the modem, feeding a D-Link DIR-1030, but the same occurred with the old rectangular Plusnet mod/router - I switched back for a week to see but have returned to the other combo as I need them.

As I said, the line is stable, the throughput not.

Line was silent, I don't know about today as I'm not there, confirmed by engineer at the original check in 2021, and by myself since.  

I can't get to the modem stats as the modem is configured as a full bridge to the router - it requests an external IP and translates it to the router transparently. In order to get further info, I'd have to go home with a laptop and hard-wire into the modem instead of the router, which won't happen for 2 weeks.

Network config:

Level 1: Vigor 130 in full bridge mode 

Level 2: DIR1030 

Level 3: 3x NETGEAR GS308 8-Port Gigabit Switch 

Level 4 Wired:

   Switch 1: 2 IP Cameras and powered off computer.

   Switch 2: DVR

   Switch 3: Computer doing this test, TV (off but connected), BluRay player (off but connected), AV Receiver (off but connected) 

Level 4 Wireless:

A few TUYA smart sockets and a smart light,

I have tried after removing Switch 1 and 2 also, same result.

Wifi doesn't disconnect, but isn't part of this test anyway.

[#9]

I will look at setting up the BQM when I next get back to the modem - I don't generally enable internet ping because I've had PoD in the past, and I need reliability when travelling away from home...

[#10]

@Coffeeking Ey up, I'm seeing 64 Mbps from my end and no drops for 19 days so there's nothing obviously wrong. I am seeing some errored seconds (Not enough write home about but could be a sign. Perhaps the error are more sever in bursts). As your sync rate is above the estimated I could change the DLM stability options to super stable if you like to hopefully prevent some of those errors I’m seeing here?  

[#11]

@James30 

The evidence to date points away from link stability and performance, historically this link is reported to have experienced 100+ days continuous PPP.  Is there any indication that the few ES recorded are indeed in concentrated bursts?  If not and the user's experience is more prevalent than a few sporadic bursts, then the issue must be elsewhere - either within the user's premises or between the exchange and wherever on the internet.

The question not responded to seeks to confirm that there is no congestion in the BT Wholesale back haul, commonly referred to as a Hot V-LAN".

@Coffeeking a TBB-BQM will indeed be helpful, as would xDSL monitoring ... but you cannot get into the router you are using.  Do you have a Plusnet Hub One (or some other modem compatible with router stats) which might be used to inspect for REIN ... which might match James' "Perhaps the error are more sever in bursts" observation.

[#12]

@Townman I have checked the Vlan sheet and it's showing green. The below information is what's raising an eyebrow;

Parameters Last Traffic Count(Upto 15 mins)      Current Traffic Count(Upto 15 mins)
Start Time Stamp 2023-05-03T14:59:32Z              2023-05-03T15:14:32Z
Ingress Code Violation        1                                       0
Egress Code Violation         12                                     7
Errored Seconds                  2                                       3
Severely Errored Seconds  0                                       0
Unavailable Seconds           0                                       0
 

[#13]

Hi Folks, thanks for taking an interest in helping - much appreciated! Slight correction to my earlier post, it's a DIR3060, not a DIR1060, apologies.

Right now I'm a bit stuck for helping further as I'm in Europe, but my data from the linux machine in the home network tells me the DL has been sat at 0.07 for the last 6 hours (using speedtest.net sampled once per hour, with the test server in the next city over). During that time, the ping has been up to 350ms but the UL speed remained as high as ever. This is performed by the linux machine.

Right now, the DL is so low I can just about establish a VPN connection (took 20 seconds) to my house, but currently it's very slow (browser time-outs, video  streaming (from the DVR) through it impossible etc). I cannot establish a TeamViewer connection to the linux machine, bypassing the VPN, because it's too slow to connect. I have logged into the router admin interface (not the modem) and it has the ability to run it's own speed test without browser interaction etc, and this reports the speed too low to detect and errors out.

When I get back to the flat this evening I'll try to get deeper access and see if there's a route through to the modem, but I'm fairly certain I've never been able to reach the admin interface from this side of the bridge.

[#14]

Poor DL data speed in the presence of good sync speed points to a high level of retransmissions.

Personally I’d be wanting to rule out REIN and that needs sight of the dynamic SNRM in the modem.

[#15]

I can't get access to the modem output without physical access so it's probably best that I hit pause on the thread until I can provide more detail. On the assumption that electrical noise would be either short term (hoovering), regular (TV PSU on at the same time each night)  or continuous (someones PV inverter spitting out noise during sunlit hours) I looked back over the last 2 weeks or so and I see no obvious patterns by time of day (blank section was a system update on the test machine). It resolved itself at midnight last night, for this round and I'm back to full speed for now!

Due to the fact that I'm not in the house (no-one is) I can only assume that it's one of very few items of my own, or it's a neighbour, or in the cabinet. This will become an almost impossible hunt, I feel.

It's worth noting that this data is the first time I've had evidence of this problem - these are the longer sessions where it's shown up on multiple hourly tests since I ran them. Prior to this, I had no data, but I saw dropouts and slow-downs many times a day, and it's possible this is just not catching those too, I just didn't want to abuse the test service by testing more frequently.