Slightly Used Internet Connection + Puma Chipset in SH3 = Poor Gaming

[simonmanch84]

I've been having really inconsistant packet latency for the last 3 months or so that seems to be getting worse. I'm on M600 with the SH3 in modem mode. I'll disclose from the start that I'm a Cisco Engineer in my line of work, and probably have an above average network setup at home. I have an enterprise grade firewall connected to a Cisco Catalyst switch, and most of my equipment including the gaming PC is hard wired. Whats happening is that my ping is nice and low, between 15/30ms depending on the server. Then out of nowhere, latency to the servers will go above 150, for a second or two and then go back to normal.

I've since installed an OpenWRT running SQM Cake to combat the obvious bufferbloat that my connection suffers with, sitting betwen my firewall and t]he Super Hub - This has brought my score on the waveform bufferbloat test (https://www.waveform.com/tools/bufferbloat) from a D down to an A/A+ depending on the time of day, but it still has this random blip every few minutes which results in either a missed shot in CS:GO or a pause and shift/jump in Rocket League. I've tried deploying QoS on my Layer 3 switch, QoS on the Firewall, disconnected all but a single PC for testing, including disabling my wireless access points, nothing seems to prevent it. I've even tried rate limiting the connection down to as low at 100mbps by manually configuring my port to be 100mbps full duplex, but nothing seems to prevent it, so I don't believe its anthing to do with a congested connection from my side.

Is this likely the rumoured bug with the Intel Puma chipset in the Super Hub 3? 

BT have just ran FTTP past my house, and i'm seriously considering switching to them when my contract expires, as this is getting beyond a joke now - you can see from many in this forum that a lot of us are having these problems and are getting fobbed off by VM. I've only stayed with VM for all these years as the competition couldnt come close to the speeds that VM could, but now that they can I'm going to need a lot of persuasion not to switch. 

What do I need to do to get a SH5? and is this any better? or is it a case that my area is just oversubscribed and / or DOCSIS3 isn't suitable for these sort of speeds?

As I speak the issue is very bad today, so bad infact I'm only getting a B grade on the waveform bufferbloat test, with the odd latency during downloat in the 400's!!!!

https://www.waveform.com/tools/bufferbloat?test-id=d10f7a79-382b-4422-a5f7-c9e8ca7f7cb9

Any ideas?

[IPFreely]

Looks as though it thinks you were making constant upstream requests so were able to skip the queue by piggybacking requests for the next grant onto the previous data transmission, bringing down your minimum latency.

Could also be a test of something involving a lower MAP interval. 

Given your lower minimum latency, the notably higher average and the maximum while this person has a 'normal' minimum, small average increase and is plagued by spikes rather than constantly increased latency it doesn't look like the same thing. 

[legacy1]

---

---

---

@IPFreely wrote:

Looks as though it thinks you were making constant upstream requests so were able to skip the queue by piggybacking requests for the next grant onto the previous data transmission, bringing down your minimum latency.

Could also be a test of something involving a lower MAP interval. 

Given your lower minimum latency, the notably higher average and the maximum while this person has a 'normal' minimum, small average increase and is plagued by spikes rather than constantly increased latency it doesn't look like the same thing. 

---

On this day 1st at 6:30pm I did a test upload to twitch at 6000 Kbps and you can see the green go up

I'm now running a test upload to twitch with this change VM have done its doing the same only it might look like the latency before...

[IPFreely]

So nothing has changed it was just traffic pattern allowing your ping responses to jump the queue as it were.

Perhaps you hit a sweet spot with the mass of telemetry your home lab is doing and there was spare room in the requested bandwidth: something unlikely with 6 Mbps of streaming given you'll make at most 250 requests for capacity per second.

Either way not related to the OP's issue and no amount of QoS is going to get him out of it. PEBMAC. 

[legacy1]

---

@IPFreely wrote:

So nothing has changed it was just traffic pattern allowing your ping responses to jump the queue as it were.

---

No and yes theirs seems to be two types upstream scheduling modes VM can do when my connection is idle

12am- 3pm type 1

3pm - 11:59pm type 2

[legacy1]

---

@IPFreely wrote:

Either way not related to the OP's issue and no amount of QoS is going to get him out of it. PEBMAC. 

---

Depends on the available downstream bandwidth if you rate limit under that when area is under heavy load then ping will be fine of course if the spikes are due to upstream under heavy load in the area it will not help.   

[simonmanch84]

Thanks guys for your comments - I can safely say that unless virgin media can see any noise or power level issues on my connection then it must mean a congestion issue in my area and that pretty much makes up my mind to switch to Openreach FTTP when my contract expires. Been with VM for the past 12 years so after years of treating existing customers as 2nd class citizens come renewal times, and now no longer providing a decent service, they will most likely be loosing a loyal customer. The speed was the only thing keeping me with them, but now thats not happening these days the only thing that can keep me is if they fix the problem and also make it cheaper.

Is there any way of flagging one of the VM team members on here to check the line and the BQM for comments?

[simonmanch84]

So, 24hrs later still no comment from Virgin Media - I'm not going to ring them for support to end up talking to their support staff whom last time I had an issue said they wouldn't look at it unless I turned off modem mode. Looks like I'll be switching to OpenReach FTTP and Sky TV.

[legacy1]

I really wish would start supporting modem mode more even if its just a PC connected its not like modem mode is a new invention its what we had years and ISP troubleshoot problems just fine and this whole we need to do tests in router mode we can't do in modem mode is rubbish 10. could have been use for that and even then if your Ethernet port is faulty VM can't check that.

As for no 24hrs later still no comment from Virgin Media it takes them longer to reply.

[simonmanch84]

I've seen quick replys to posts in here before - just seems I'm being ignored.

Last night the connection was terrible, at one point I was only getting 50mbps download!!!

[Vikki_M]

Hi simonmanch84 

Thank you for your post and welcome to our community.

Apologies for the delay in getting back to you. 

We would be happy to take a look at things on our side for you.

We are however unable to comment on any 3rd party equipment. 

Please put the VM Hub back in Router mode for at least 24 hours and then we can run accurate checks for you. 

Please pop back to us when you can.