Performance issues Doc 3 bonding | docsis.org

You are here

Performance issues Doc 3 bonding

11 posts / 0 new
Last post
windwaterwaves
Performance issues Doc 3 bonding

So I just created a problem for myself the other day. We dropped our analog television service. So I shifted 4 channels from the 500mhz region down into the 200mhz region, saw the modems come back and no problem, less swing staying out of the high end so should be great right?

Nope. Customer complaints, packet loss, great upstream speed, overloaded ds (lightly loaded node). Found using show wideband-cable 8/0/2:0 rf-status that 5 out of the 8 downstreams were down... went into controller and had to rf-shutdown each one and no rf.. etc. They all say up now, but load balancing is still weird and off. Looking at the integrated cable interfaces 8/0/2:0 is full, 8/0/2:1 has 72bps, 8/0/2:2 has 12mb. wth?

Is there someway to troubleshoot? Are we getting errors in the DS on the modems (I don't have any tools to find that, I will have to write a snmp script to grab all the ds channel info)

mbowe
Are you sure you removed all

Are you sure you removed all the TV signals? I would be plugging a spectrum analyser into the headend transmitter testpoint, and having a look to see if the CMTS channels are showing up cleanly there. Possibly even quickly unplug the CMTS channels temporarily and check that spectrum does not have any other signals showing.

Do you have a stealth/sweep system, have you reprogrammed the channel plan there? I once made the mistake of forgetting to update this, and the insertion points were causing havoc with newly activated DS channels.

Beware the total input power levels at your transmitters. If you remove a lot of analog channels, then the lasers may now have insufficient input power level. Check the documentation. Also look at the GUI for the lasers, see if there are alarms showing.

Log into customer modem GUI and have a look at the DS SNR, correctables, uncorrectables. Or poll it out using SNMP.

windwaterwaves
Possibly the stealth sweep,

Possibly the stealth sweep, your right, I haven't reprogrammed it, but I thought it was disabled, I will check today. I did check the background noise and was clean. Gui for lasers hahahaha.. Just about to upgrade nodes to harmonic or aurora, thoughts?

Polling modems now but only single channel so I need to change that.

windwaterwaves
I was able to use show cable

I was able to use show cable modem cable 8/0/0 partial-service to identify one bad downstream channel, muted it and voila back to full speed. I will be re-aligning frequencies this week. Any frequencies to avoid, favorite frequencies.. albeit you are in AU so RF regs are different but just curios as to cell towers etc.

schg
You can configure ds partial

You can configure ds partial bonding and ds-resiliency bonding to avoid these kind of problems.
The CMTS will move the modems away from the impaired ds-channels.
If the whole segment is affected the CMTS also might be configured in a way that the CMTS will entirely disable the impaired channel.

windwaterwaves
Any suggested config's?

Any suggested config's?

schg
example

Enable Resiliency globally:

cable resiliency ds-bonding
cable rf-change-trigger percent 50 count 10 secondary
cable wideband auto-reset

for each Linecard you need to setup several resiliency bonding interfaces, which will be used by the CMTS to create custom bonding-groups for the impaired modems:
(I typically use the last MAC-Domain on the Linecard, because it is unused anyways),


interface Wideband-Cable5/0/4:0
description DS-Resiliency
no logging event link-status
cable ds-resiliency
no snmp trap link-status
!
interface Wideband-Cable5/0/4:1
description DS-Resiliency
no logging event link-status
cable ds-resiliency
no snmp trap link-status
!
interface Wideband-Cable5/0/4:2
description DS-Resiliency
no logging event link-status
cable ds-resiliency
no snmp trap link-status
!
interface Wideband-Cable5/0/4:3
description DS-Resiliency
no logging event link-status
cable ds-resiliency
no snmp trap link-status
!
interface Wideband-Cable5/0/4:4
description DS-Resiliency
no logging event link-status
cable ds-resiliency
no snmp trap link-status
!
interface Wideband-Cable5/0/4:5
description DS-Resiliency
no logging event link-status
cable ds-resiliency
no snmp trap link-status

Ensure, that you did not disable CM-Status messages.

mbowe
I have experienced some

I have experienced some ingress issues with frequencies below 200.

Watch out for any ALC pilot frequencies, eg we have some at 439.25, 427.25.

Best to stay clear of frequencies 700+ (LTE ingress).

I reckon the sweet spot is around 300's. We have been successfully using a 32 channel block from 237 to 423.

mbowe
Highly recommend you ensure

Highly recommend you ensure your CMTS channels are contiguous.

Don't have any gaps.

Or at minimum, ensure you are allocating contiguous blocks of 8.

windwaterwaves
Prior to now we had 4 in the

Prior to now we had 4 in the fm band and 4 at 450, a couple of nodes had the 4 fm band and 8 at 450. I will get them sequential. It was all working pretty well. Need to get rid of fm band but there are window filters hiding out there still.. so that may take a while.

schg
It is not absolutely

It is not absolutely necessary to make your downstream channels line up seamlessly.
But you really have to have a look at your modems tuner capabilities and your wideband interfaces.

There are plenty of different tuner combinations out there. Especially for 8x bonding capable modems.
I only can speak for EuroDOCSIS. This is my experience with plenty of modems (from worst to best):
a) 1x Tuner with 64 MHz capture window and 4xbonding
b) 1x Tuner with 96 Mhz capture window and 8xbonding
c) 2x Tuner with 32 MHz capture window and 8xbonding
d) 2x Tuner with 64 MHz capture window and 8xbonding
e) 2x Tuner with 96 MHz capture window and 8xbonding
f) 1x Tuner with Fullband capture window and 8xbonding

essentially this sums up to the following criterias:
- make any four channels contigous (requirement of tuner c). Obvious, since ubr10k forces you to do so.
- create 8x-bonding wideband-interfaces only for channels in 2x4 blocks with a max window of 96 mhz (requirement of tuner b)

this means for 16 channels you might get away with a configuration, that has eight channels in the lower part of the spectrum and eight channels in the higher part of the spectrum. Ensure you place some primaries in both.
Then create 8x wideband interfaces for the lower and higher parts individually.

We use this quite sucessfully without any issues.

Log in or register to post comments