50M : Why the bad connection?

[ianch99]

I need help in understand why my 50M is so poor regarding latency and jitter and to a lesser extent speed. I am on the sotn8-2-0 CMTS/UBR at the moment and here is a snapshot of my current BBM graph:

My CMTS line: cpc3-sotn8-2-0-cust***.15-1.cable.virginmedia.com

Snapshot graph:

http://www.thinkbroadband.com/ping/s...8-07-2012.html

and a typical pingtest.net result is:

http://www.pingtest.net/result/67084269.png

This is using a WNR2000v1 router together with my VMNG300 modem. I normally use a DIR-615 running dd-wrt. The DIR-615 shows the same graph profile so I switched to the Netgear to rule out the dd-wrt firmware.

The thing that makes this mysterious is that my neighbour is also on the sotn8-2-0 CMTS (10M product/Media Hub/DOCSIS 2.0) and his graph is totally different (and better!):

His CMTS line: cpc8-sotn8-2-0-cust***.15-1.cable.virginmedia.com

Snapshot graph:

http://www.thinkbroadband.com/ping/s...8-07-2012.html

and a pingtest.net done over wireless at long range(!):

http://www.pingtest.net/result/67084046.png

Same CMTS, same street cab but totally different behaviour. The only conclusion is that my modem is faulty and I would need to switch to the SuperHub sitting in the cupboard ...

Any advice?

BTW, here are my current modem stats:

Upstream

Power Level (dBmV) 46.25

Downstream

Power Level
(dBmV) 0.70 0.50 0.69 -0.36
RxMER (dB) 37.09 36.84 37.36 37.09
Correctable
Codewords 194 234 179 247
Uncorrectable
Codewords 290 300 291 295

[General Maximus]

no way dude. They'll want to swap the vmng300 over for the shub before they look at anything else so you are screwed anyways but I defo font believe it is the modem. Thetap point you are coming off in the cabinet could be corroded etc which can cause problems. Someone else has said the same thing this week; he was screwed for ages and engineer came round and moved to a diagnostic/testing tap point and everything was okay.

Your power levels are fine but the number or uncorrectable codewords indicates a network issue and not modem issue.

[hicks]

Any T3/T4 timeouts showing in your modem config page? Your power levels are good, but that TBB graph is horrible. How long since modem was rebooted - the correctable/uncorrectables reset after a reboot.

[Skie]

Your neighbour is on 10meg, which uses the legacy DOCSIS 2 while you are on DOCSIS 3.

In essence they are separate networks running over the same lines, but quite often you see someone go from 20meg or less to 50/60 and find that their connection gets much worse as they join all the heavy downloaders in their area. So its most likely over utilization.

Have a moan on the community forums, they are usually much better about acknowledging that issue than India. Of course, it will take 6 months to fix if you are lucky.

[jb66]

Originally Posted by Skie

Your neighbour is on 10meg, which uses the legacy DOCSIS 2 while you are on DOCSIS 3.

In essence they are separate networks running over the same lines, but quite often you see someone go from 20meg or less to 50/60 and find that their connection gets much worse as they join all the heavy downloaders in their area. So its most likely over utilization.

Have a moan on the community forums, they are usually much better about acknowledging that issue than India. Of course, it will take 6 months to fix if you are lucky.

Spot on

[ianch99]

Originally Posted by hicks

Any T3/T4 timeouts showing in your modem config page? Your power levels are good, but that TBB graph is horrible. How long since modem was rebooted - the correctable/uncorrectables reset after a reboot.

A couple of T3's in the log since the 25th, no T4's. I have rebooted the modem a few times ... The correctable number are about right with those you get after a reboot and they are not increasing.

I tend to agree with the over-utilised CMTS theory which is what my first thoughts were ... BUT ... tell me this: how can a DOCSIS2.0 line into the same CMTS be fine but a DOCSIS3.0 line (mine) into the same CMTS be so over utilised?

I have seen the VM postings when a "good" 10/20M line becomes a horrible 30/50/60M one but I had assumed that they had moved from a under-utilised CMTS to an over-utilised one .. maybe this is not the case?

If anyone knows how these CMTS boxes work: can you get line prioritisation with a single CMTS? Or maybe the VM hostname name assignment (from which we deduce the CMTS we are on) are mis-leading .. who knows ..