Upload Speed Off Since Install

[Ignitionnet]

Collective,

Wonder if you have any thoughts.

I've had slower than expected upload speed since install, very rarely did it get to 1.5Mb/s and is more commonly at 1.2Mb - 1.3Mb.

Any suggestions welcome.

Downstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power SNR Docsis/EuroDocsis locked
Locked QAM256 38 55616000 Kbits/sec 299000000 Hz 12.0 dBmV 39.8 dB Hybrid
Locked QAM256 39 55616000 Kbits/sec 307000000 Hz 11.5 dBmV 39.9 dB Hybrid
Locked QAM256 40 55616000 Kbits/sec 315000000 Hz 11.2 dBmV 39.3 dB Hybrid
Locked QAM256 41 55616000 Kbits/sec 323000000 Hz 10.7 dBmV 39.0 dB Hybrid


Upstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power
Locked TDMA 2 10240 Kbits/sec 25800000 Hz 34.0 dBmV

Download Speed: 50432 kbps (6304 KB/sec ) Upload Speed: 1266 kbps (158.3 KB/sec )

Download Failed (1)

https://www.cableforum.co.uk/images/local/2011/08/14.png

Download Failed (1)

[Taf]

I was stuck at 0.2 mbps upload since moving up to the 30mbps service. A tech got it up to around 1mbps by adding an attenuator and replacing the isolator, but that's still a long way off the 3mbps it should be. The tech says an update in the pipeline could fix things....

[jb66]

Are you in an upgraded area? That speeds what I'd expect from an old area. Also your line needs attenuated by 10db

[Ignitionnet]

Are you in an upgraded area? That speeds what I'd expect from an old area. Also your line needs attenuated by 10db

The area isn't upgraded.

Power levels were previously fine, some work was done on network which kicked downstream levels up by 8dBmV and upstream transmit power up by 15-17dBmV.

Speeds aren't hugely low but aren't the ~1.6Mb/s that should be reachable.

[TJS]

Collective,

Wonder if you have any thoughts.

I've had slower than expected upload speed since install, very rarely did it get to 1.5Mb/s and is more commonly at 1.2Mb - 1.3Mb.

Any suggestions welcome.

Downstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power SNR Docsis/EuroDocsis locked
Locked QAM256 38 55616000 Kbits/sec 299000000 Hz 12.0 dBmV 39.8 dB Hybrid
Locked QAM256 39 55616000 Kbits/sec 307000000 Hz 11.5 dBmV 39.9 dB Hybrid
Locked QAM256 40 55616000 Kbits/sec 315000000 Hz 11.2 dBmV 39.3 dB Hybrid
Locked QAM256 41 55616000 Kbits/sec 323000000 Hz 10.7 dBmV 39.0 dB Hybrid


Upstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power
Locked TDMA 2 10240 Kbits/sec 25800000 Hz 34.0 dBmV

Download Speed: 50432 kbps (6304 KB/sec ) Upload Speed: 1266 kbps (158.3 KB/sec )

http://www.pingtest.net/result/45901580.png

https://www.cableforum.co.uk/images/local/2011/08/14.png

http://www.thinkbroadband.com/ping/share-thumb/d0bd3309e6a6866885f308c86b43ebfd-26-08-2011.png

It's called overheads. E.g. my 5 mb/s upload is usually 4.7 mb/s

https://www.cableforum.co.uk/images/local/2011/08/15.png (http://www.speedtest.net)

[Ignitionnet]

XXL's upstream before the uplift was always capped at 1.75Mb/s, after overheads should be around the 1.6Mb/s mark.

Primary Upstream Service Flow
Upstream(0)
SFID 1804
Max Traffic Rate 1750000 bps

[qasdfdsaq]

It's called overheads. E.g. my 5 mb/s upload is usually 4.7 mb/s

Fail.

[Ignitionnet]

http://mcslhr.visualware.com/myspeed/db/report?id=1288506

D/load capacity 48656 Kbps
U/load capacity 1019 Kbps

VM's own speedtest:

Mon Aug 29 2011 21:49:15 GMT+0100 (GMT Daylight Time)
Last Result:
Download Speed: 49290 kbps (6161.3 KB/sec transfer rate)
Upload Speed: 1154 kbps (144.3 KB/sec transfer rate)
Latency: 20 ms

http://www.thinkbroadband.com/ping/share-thumb/70a12d2cc2bfd2c9d220f401fd81c7b1-29-08-2011.png

Looks like it's a load thing, doesn't explain why it only manages 1.4Mb or less even during the periods of low load though.

I'm rather hoping those guys will see something more obviously wrong somewhere along the chain, load going back up as people come back from summer holidays, not ideal, and was playing hell with my VoIP between 5PM and 5:30PM today.

High power levels persist:

Downstream Channels

Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power SNR Docsis/EuroDocsis locked
Locked QAM256 38 55616000 Kbits/sec 299000000 Hz 12.0 dBmV 39.9 dB Hybrid
Locked QAM256 39 55616000 Kbits/sec 307000000 Hz 11.8 dBmV 39.7 dB Hybrid
Locked QAM256 40 55616000 Kbits/sec 315000000 Hz 11.3 dBmV 39.7 dB Hybrid
Locked QAM256 41 55616000 Kbits/sec 323000000 Hz 10.9 dBmV 39.3 dB Hybrid

As do low ones:

Upstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power
Locked TDMA 2 10240 Kbits/sec 25800000 Hz 34.0 dBmV

The jitter is back into thoroughly service affecting range - VoIP client was having to buffer for 100ms to try and mitigate the connection, causing delay, and even then failed to completely mitigate it so some parts of the conversation were punctuated by garble.

http://www.pingtest.net/result/46067776.png

[Sephiroth]

If they did something to the street cabinet during the local upgrade (did they?) then you're prolly on an inappropriate tap point.

What's your maximum upstream traffic rate? i.e. are you on QPSK? If you are, then that and low upstream power might make the upstream difficult. On the other hand I would expect to see more packet loss on the TBB graph.

Am I anywhere near the right direction or should I have kept quiet like most do when you're asking?!

[jb66]

Unusual to get moved up :)

[Sephiroth]

Not if they ripped it all out, replaced the guts and put it all back together. Didn't Igni's area get some sort of refurbishment? I forget if it went to the cabs.

[Ignitionnet]

If they did something to the street cabinet during the local upgrade (did they?) then you're prolly on an inappropriate tap point.

What's your maximum upstream traffic rate? i.e. are you on QPSK? If you are, then that and low upstream power might make the upstream difficult. On the other hand I would expect to see more packet loss on the TBB graph.

Am I anywhere near the right direction or should I have kept quiet like most do when you're asking?!

In paragraphs:

1) No - they were messing with amplifiers.
2) This area does not have ASM enabled. Upstream is locked at 16QAM 3.2MHz.
3) I would be open to anyone's suggestions.

Tech support inform there are no load issues or visible issues, however this is something that's never been right. I'll see what if anything changes when they start the software upgrades.

Since the network work was done I've periodically been seeing this, was never there before:

Sat Sep 03 20:26:54 2011 Critical (3) UCD invalid or channel unusable;CMTS-MAC=00:30:b8:d0:ac:c0;CM-QOS=1.1;CM-VER=3.0;

Before a moderator descends CMTS MAC doesn't matter. Actually neither does CM MAC on Superhubs.

[Sephiroth]

If the upstream is locked at 16QAM what did the techs tell you about your upstream SNR?

[Ignitionnet]

That it was good and had been for some time.

If there were serious SNR problems they would, as you mentioned, manifest as packet loss.

I personally think it's a combination of load and burst retransmits, of course I absolutely cannot prove that without both access to the CMTS and access to detailed diagnostics on retransmits from my shrub.

[qasdfdsaq]

Stating the obvious here, but what difference do you see between peak and off-peak times?

Now that I think about it, I realise my connection might have been doing something similar for the past few months (and my TBB graph now looks like yours) - would explain why I only get ~1-1.2mbps during the evenings, but right now it's getting a decent 1.6 up.

[Ignitionnet]

Yup it's fine in the mornings, very occasionally it gets to 1.6Mb/s, it slows down in the evenings.

I can, however, get a full 1.6Mb/s in the evening by using a lot of TCP connections, IE through Bittorrent, and essentially crowding out other people.

This is a bit odd for DOCSIS behaviour, I'm wondering if the BSR is applying some layer 3/4 management to try and avoid DOCSIS congestion.

BTW I didn't realise about the lot of connections, I thought I'd test the upstream given I was assured load was fine and did so with a torrent.

[qasdfdsaq]

This is a bit odd for DOCSIS behaviour, I'm wondering if the BSR is applying some layer 3/4 management to try and avoid DOCSIS congestion.
There could be some truth to that. I've not got enough congestion right now to say, but I have noticed on my TB ping graphs that the tops of the chart do stay a lot lower now - and jitter is somewhat less.

[Ignitionnet]

Just to close this off:

07-09-2011 13:29

Hi Ignition,

I can see that there is high utilisation on your local UBR and I have escalated this to our network team to investigate further.

In addition to this both your upstream and downstream power levels are outside of the acceptable limits and would advise having them alter by an engineer.

Please accept our apologies for the inconvenience caused.

Regards,

James Gilbert

Help & Support Forum Team

[qasdfdsaq]

Pfft... So much for no utilization issues (I got the same crap off telephone support)

[Sephiroth]

The term "high utilisation" in this case is most likely VM's euphemism for the consequences of running their upstream at QPSK. As you know, QPSK halves the available upstream capacity available to the customer base.

[Ignitionnet]

The term "high utilisation" in this case is most likely VM's euphemism for the consequences of running their upstream at QPSK. As you know, QPSK halves the available upstream capacity available to the customer base.

ASM isn't active on my CMTS, 16QAM or bust.

The term in this case is a euphemism for having 260 modems on a single 10.24Mbit upstream port - doesn't take much to tip it over. ;)

[Sephiroth]

You've said that before, Igni. But I was addressing the use of euphemism by the VM tech seeing as you're on QPSK - ASM or not.

[Ignitionnet]

You've said that before, Igni. But I was addressing the use of euphemism by the VM tech seeing as you're on QPSK - ASM or not.

I'm not on QPSK - note TDMA rather than ATDMA, 3.2MHz channel width not 6.4.

[Ignitionnet]

This should be resolved shortly. There were 3 separate issues.

The upstream channel had too many modems connected, some 260, causing overutilisation of underlying DOCSIS resources.

The upstream at times had bandwidth overutilisation (over 10% of the time at over 80% utilisation)

Inconsistent upstream SNR from my modem due to low transmit power.

The resource issues are being dealt with through addition of a 2nd DOCSIS 1.1 upstream, the signal problem with a bidirectional attenuator.

Case closed :-)

[qasdfdsaq]

Or so you think. Until next time... ;)

[Ignitionnet]

Or so you think. Until next time... ;)

Indeed.

For the curious as I mentioned the upgrade involves no new hardware. It was rather being covered up by the fact that usage per modem is relatively low in this area, and that the Motorola BSR's polling, granting and traffic grooming algorithms appear to handle high upstream load far better than Cisco's. Even at well over 80% utilisation disruption was relatively minor.

The actual upgrade has to all be done using the existing frequency plan in the area for now, however due to recent network upgrades there's some capacity free - VM can run the existing channels at higher capacity than they could before.

VM currently have 4 x 3.2MHz usable channels upstream in this area. There are tons more now but these aren't getting activated just yet.

The DOCSIS 1.1 network has 3 x 5.12Mbit upstreams, the DOCSIS 3 network has 1 x 10.24Mbit upstream.

The upgrade that they'll probably do is to take a channel from the DOCSIS 1.1 network and double its capacity by moving it from QPSK to 16QAM, leaving the DOCSIS 3 network with 2 x 10.24Mbit upstream channels.

The remaining DOCSIS 1.1 channels will also have this capacity increase, doubling the upstream capacity of the DOCSIS 3 network and increasing the capacity of the DOCSIS 1.1 network by a third with no new hardware and some configuration and a bit of wiring in the hub site.

As some may know the original Motorola BSR release only used half of the upstream ports on the card due to firmware restrictions, these restrictions are gone but the ports haven't been used yet, so there's half a card of upstream ports free on the equipment my DOCSIS 3 modem connects to.

The good part of this upgrade also is that 100Mb requires at least 2 upstream channels, so when it comes time to release the upstream upgrade they'll just change the configuration on those 2 channels rather than needing to do rewiring, and 100Mb enabling will just be business as usual capacity upgrades as required.