Recent peak-time speed problems (Nov 08) - Please read this!

[#1]

I believe http://www.mybroadbandspeed.co.uk/ is in London.

[#2]

Yes a tracert would seem to confirm this (was reached in 5 hops whereas community reached in 7 and portal in 10)
Hopefully they can tweak the traffic management or add additional processing capacity to the routers if the links are within reasonable usage levels.
As for the BBC streaming, might it be a good idea to give those that want to use that a separate ID to allow traffic steering onto a single edge router and free up the rest of the network?
Just an idea.

[#3]

I am using BBYW option 2 and carried out some speed tests this evening in anticipation of taking part in the ADSL2+ trial.  My line is stable and syncs at 8096 kbps downstream and 448kbps upstream.  The IP profile for the line is 7000 kbps.  The speed test results are below:
Tests using the BT Speedtester (www.speedtester.bt.com):
These results are very poor.  Does the BT Speedtester use FTP downloads I wonder?  - Does that explain the poor results? 
(BT speed tests taken this time last year were very good - consistently achieving actual IP throughput rates of about 5800kbps)
24/11/2007 22:30
Test1 comprises of Best Effort Test:  -provides background information.
    Your DSL connection rate: 8096 kbps(DOWN-STREAM),  448 kbps(UP-STREAM)
    IP profile for your line is - 7000 kbps
    Actual IP throughput achieved during the test was - 10 kbps
24/11/2007 22:35
Test1 comprises of Best Effort Test:  -provides background information.
    Your DSL connection rate: 8096 kbps(DOWN-STREAM),  448 kbps(UP-STREAM)
    IP profile for your line is - 7000 kbps
    Actual IP throughput achieved during the test was - 42 kbps
24/11/2007 22:40
Test1 comprises of Best Effort Test:  -provides background information.
    Your DSL connection rate: 8096 kbps(DOWN-STREAM),  448 kbps(UP-STREAM)
    IP profile for your line is - 7000 kbps
    Actual IP throughput achieved during the test was - 409 kbps
Tests using www.mybroadbandspeed.co.uk:
These have varied widely with some extremely high figures!
<-- very odd - all of my readings between 21:00 and 22:05 were this high
<-- ditto
<-- back to normal readings


One additional observation:
I did a large VMware download (553MB) on 22/11 at about 00:30 and that was classified as 'Streaming' on the 'View My Broadband Usage' page.  This seems to have been classified incorrectly.  Is it possible that other types of traffic are being incorrectly identified and throttled as a result?  Just a thought!

[#4]

Quote from: pnf

One additional observation: I did a large VMware download (553MB) on 22/11 at about 00:30 and that was classified as 'Streaming' on the 'View My Broadband Usage' page.  This seems to have been classified incorrectly.  Is it possible that other types of traffic are being incorrectly identified and throttled as a result?  Just a thought!

Hi,  it would be great if you could provide a wireshark capture of this VMware download so we can understand where the mis-identification is happening, thanks.

[#5]

Quote from: techguy

I am noticing variable performance on some sites too, I think PN need to add some more transit

Hi, Have you looked at http://www.plus.net/features/network/technical_info.shtml
We have plenty of spare capacity on all peering and transit links.  FYI our total external traffic is just over 6Gb/s at peak.

Quote from: techguy

I'll post some tracerts as the PN speed test can suffer from loading on the links back to Sheffield (I'm assuming this is where the test server is) whereas the normal traffic does not take this route

The speedtest servers are located in Telehouse Europe North and the links used for our Sheffield to London traffic are each less than 50% utilized.

Quote from: techguy

or add additional processing capacity to the routers if the links are within reasonable usage levels

From the link provided above you can see we use carrier class routing equipment, our core and edge devices range from 10Gbps to 720Gbps switch fabrics achieved by hardware assisted forwarding techniques

Quote from: techguy

As for the BBC streaming, might it be a good idea to give those that want to use that a separate ID to allow traffic steering onto a single edge router and free up the rest of the network

As the points in the network where congestion can and do occur are the 622Mb/s L2TP Edgeless IPStream centrals we cannot steer some types of traffic to end-users down one central and other types down a different central.
HTH
Carl

[#6]

Thanks for the info Carl
So the migration to 21CN should allow you to scale the incoming bandwidth to cope with the demand then?

[#7]

Quote from: techguy

So the migration to 21CN should allow you to scale the incoming bandwidth to cope with the demand then?

Yes, but then the current network allows us to do that too.
Phil

[#8]

If my understanding of the info provided is correct it shortens provisioning time.

[#9]

Not necessarily afaik.
To install a new central or a new MSIL takes an awfully long time.  I think it's 49 working days on a central, but not 100% sure on an MSIL.
Lighting a segment or opening up more capacity on an MSIL I believe are both 5 working days.

[#10]

Quote from: _CN_

Hi,  it would be great if you could provide a wireshark capture of this VMware download so we can understand where the mis-identification is happening, thanks.

I did a partial download a short while ago after installing Wireshark.  The download was an http download.  The transfer took place between the source port: http(80) on the VMware server to the destination port: isoipsigport-2 (1107) on my PC.  I will repeat the full download during the free period later as it should be fairly easy for me to spot how it gets classified in the VMBU during that period.  I will be happy to PM a Wireshark capture file to you for further analysis if you need more detailed information. 

[#11]

excuse my ignorance but an MSIL is?

[#12]

@Techguy
I'd have a read of this:
http://www.samknows.com/broadband/21cn_broadband.php
and
http://www.btwholesale.com/pages/static/Products/Interconnect/MSIL_Ethernet.html
@ pnf
You could have a look at this site, to show you how to interpret the wireshark data amongst our traffic management rules:
http://www.plus.net/support/broadband/troubleshooting/wireshark.shtml#analysing

[#13]

Ah now I understand, cheers

[#14]

Thanks James,  I'll take a look.

[#15]

I've read the information provided about Wireshark.
The download was correctly classified as Gold, and the Wireshark capture confirms this.  Is there anything in the capture file that would indicate why this download was classified as 'Streaming' in the 'View My Broadband Usage' analysis?