Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: jaydub on December 21, 2016, 09:38:11 AM
-
Having had some interesting issues with a Sky ADSL line, I migrated to a Pulse8 76/18 FTTC package about six weeks ago. The FTTC cabinet is the other side of our front garden wall, so we are on a full 80/20 sync with no interleaving.
Since then I have been running ThinkBroadband speed tests and get a significant difference between the single thread and multi thread speed test results, with typically a single thread speed of around 40-45 Mbps and a multithread speed of 68-72 Mbps:
http://www.thinkbroadband.com/speedtest/results.html?id=148231153859403321549
The usual thoughts on this sort of speed difference is that it might be congestion or traffic management.
However, the speed tests are pretty consistent day and night so congestion seems unlikely. Traffic management is a possibility but as far as I know Pulse8 are just reselling a TTB connection (the line is recognised as a TTB line on speed tests) and there seems to be plenty of TTB customers that get their single and multi thread results coincident.
However to try and rule either congestion or traffic management out, I have upgraded from a Home to a Business connection. This has made no discernible difference to the speed test results.
I have also borrowed some other VDSL kit (modem / router) to try and rule out my ZyXel VMG8924 being the root cause and that gave slightly worse results.
I have also validated the single thread results by downloading the 512MB TBB test file, which downloads in 90-100s so supporting the 40Mbps results seen in the speed test.
Pulse8's (actually TTB's) speed test usually gives me a whopping 72 Mbps download speed, but that is a multithread test on port 8080.
The TTB quality monitoring tool also doesn't give me any major concerns:
http://www.thinkbroadband.com/ping/share/a367401607150ffbaf6e453b9f2bc163-21-12-2016.html
With the exception of an interesting yellow burst at 1am this morning, all the other peaks are attributable to various speed tests. However there is no significant packet loss except occasionally on speed tests.
I would welcome any ideas you might have about the single vs multi thread speed differences, as I need to make a decision about what to do next.
Choices are: remain with Pulse8 and downgrade back to either a 76 or 30 Home package or to look elsewhere.
Many thanks for your help.
-
Reduced single thread speed is how TTB seem to manage traffic through potentially busy connections. In contrast to BT wholesale, TTB pings stay stable but single thread suffers. It happens all the time though some parts of their network not just at peak time but if there is extra congestion it will be worse. I believe all TT is the same but what you get depends on where you are located and the network parts your exposed to.
-
Reduced single thread speed is how TTB seem to manage traffic through potentially busy connections. In contrast to BT wholesale, TTB pings stay stable but single thread suffers. It happens all the time though some parts of their network not just at peak time but if there is extra congestion it will be worse. I believe all TT is the same but what you get depends on where you are located and the network parts your exposed to.
Interesting. That's a new one on me. How did you find out about this?
It does slightly beg the question of what is the point in paying for a higher priority connection if it doesn't give you any greater traffic prioritisation?
-
Experience with the same issue while a direct TTB customer and persisting in asking about the issue. The bother is that they simply don't count the issue unless it is so bad that their speed tester gives poor results. To be fair it was seldom bad enough to bother me.
-
Are there any Andrews & Arnold users out there who are experiencing this? (Many AA customers are on rebadged TTB lines.)
-
To be clear with me the issue only applied to TTB FTTC and not to the TTB adsl product.
-
Though probably unconnected following My isp Zen migrating my circuit to their GEA backhaul from the BT wbmc that i had been on since i joined them over 12mths ago, But following this change i started seeing poor single thread performance on all speedtesters that have singled thread testing, as well as actual downloading of files for any source, typically it would be around 50% of what the multithreaded throughput was sometimes it would peak at or near the max then drop to a adsl2+ type speed range or it would be similar to your speedtest result
Whilst on Zen's GEA the routeing was different i got routed to their London LNS gateways instead of their Manchester kit
And as they had also re-enabled IPV6 i was only able to connect to the one gateway, So matter what i did, such as connect the hg612 directly to my PC and use that for PPPOE sessions, this also ruled out my router, It make no difference at all
Any how after being migrated back onto WBMC platform again, i requested that they disable IPV6 again, as although it worked it didn't work as it should, i then re-tested just the modem to pc same results, then i reconnected my Router, and it gave full throughput on single thread, after further testing i realised that it wasn't my equipment but something zen's side because one lns gateway it's fine but the other i see the poor s thread issues, Zen have checked the SVLAN ect, all ok, and there is no other customer on the gateway reporting a similar thing , so is it some sort of config/setup issue with my account on Zen's systems?
-
There are quite a few people on PlusNet having the same issue, there is a 10 page thread here (https://community.plus.net/t5/Fibre-Broadband/Odd-TBB-test/td-p/1393324)
-
There are quite a few people on PlusNet having the same issue, there is a 10 page thread here (https://community.plus.net/t5/Fibre-Broadband/Odd-TBB-test/td-p/1393324)
Interesting read, Iain. However I suspect that root causes are significantly different especially as Plusnet are on BTW and Pulse8 are on TTB. If there were the same number of TTB or Pulse8 users impacted, then we would have a much longer thread of our own.
I know of one other Pulse8 user who is experiencing something similar and it sounds as though les-70 has experienced something similar on TTB.
I am seriously thinking of downgrading my Pulse8 connection to their 38/10 product and biding my time until Uno eventually launch their Talk Surf Fibre package.
-
I cannot post on the PN forums due to their lost password function been broken, but that issue they have is bizarre and the only guess I have for the PN issue is malfunctioning shaping equipment as congestion alone would not create a x1 speed of 1mbit/sec and then x6 speed of 70mbit/sec.
-
Having tried the 76 Business connection and seen no improvement over the 76 Home connection, I have decided to try downgrading to the 38 Home product.
Still seeing similar variations in the single thread vs multi thread performance.
Non-flash version:
http://www.thinkbroadband.com/speedtest/results.html?id=1483092494371812755
Flash version:
http://www.thinkbroadband.com/speedtest/results.html?id=148309235419345627959
TestMy.Net single thread results averaging out about 35Mbps though, which doesn't align very well.
Just tried downloading files and TBB and Speedtest.tele2.net results similar to the TBB tests, but web4host.net and test.unet.nl similar to the TestMy.Net results.
Slightly confused about all this. Peering link performance issues?
-
Experience with the same issue while a direct TTB customer and persisting in asking about the issue. The bother is that they simply don't count the issue unless it is so bad that their speed tester gives poor results. To be fair it was seldom bad enough to bother me.
Which ISP and product did you migrate to and did the single thread performance issues disappear on migration.
There is half of me that just says get on with life and just use the connection, but my scientific background wants to get to the bottom of what is causing the single vs multi thread performance differences
-
Having just realised that DSLStats will run under Wine on a MAC, I've got my stats uploading onto MyDSLWeb Stats (as jaydub).
Looks reasonably OK as far as I can tell, apart from zero bits on the U1 band.
However I would appreciate the thoughts of those far more knowledgeable than me on these things.
Thanks.
-
Hi
I'm on Talk Talk Business via Uno (have been for years at two different parts of the country), at some point the connection is handed over to Daisy looking at the trace root though. I've never seen any slow down at all on single threaded tests unless I'm on Wi-Fi on my PC (laptop is fine on Wi-Fi), which will be a driver issue/restriction so important to test using a wired connection, even if it seems Wi-Fi should cope with a 70+ as that is what you see on the multi-thread test. Have you tried different computers just to rule out some oddity with your network connection/setup on that computer?
It would be hard to imagine why Pulse8/TalkTalk would be restricting the bandwidth 24/7.
Regards
Phil
-
Hi
Edit: Ignore below, just seen you downgraded your product. There is nothing wrong with your sync (between you and the FTTC cabinet) to explain the differences between single and multi-thread speeds.
Ignore:
Looking at your dsl stats, you are only sync'd at 40Meg anyway, so your maximum speed can only be 40Meg at any time. Have you got a 80Meg product from Pulse8 or their 40Meg product?
Regards
Phil
-
Hi
Edit: Ignore below, just seen you downgraded your product. There is nothing wrong with your sync (between you and the FTTC cabinet) to explain the differences between single and multi-thread speeds.
Ignore:
Looking at your dsl stats, you are only sync'd at 40Meg anyway, so your maximum speed can only be 40Meg at any time. Have you got a 80Meg product from Pulse8 or their 40Meg product?
Regards
Phil
Hi Phil,
No worries.
I'm testing on wired connections. Normally on Chrome on an iMac, but have also tested using FF and Safari with very similar results.
I also have IE11 and FF on a work Win 8.1 laptop, which gives similar, but not identical results.
Stopping the ESET AV on the iMac doesn't materially improve things either. The laptop has McAfee on, but think the build prevents me from stopping it!!
Good to hear that a Uno TTB connection works, as that is probably my cheapest bolt hole should they ever make their Talk Surf Fibre package available.
I have an an ongoing dialogue with Pulse8, who have escalated to TTB. Would you mind posting your TBB speed test results, as I would like to forward it onto P8/TTB to support the fact that I should be getting full speed on the single thread test.
Thanks.
-
Hi
Here is a speedtest done just now.
http://www.thinkbroadband.com/speedtest/results.html?id=148406731205362985585
Regards
Phil
-
Here is a speedtest done just now.
http://www.thinkbroadband.com/speedtest/results.html?id=148406731205362985585
Thanks, Phil. I like that one. ;)
-
There's definitely congestion of some sort on the TTB network.
http://www.thinkbroadband.com/speedtest/results.html?id=148408595415694581972
Also gaming feels noticeably less responsive at peak periods.
The congestion isn't hard to spot on the BQM.
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2Fe7d78b1dfbb9c49a7a8c0a9d046c5ba6-11-01-2017.png&hash=e38655c1c71cba0ff99c5bf5a62d640256d971a3)
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F241d1eee6830af700e0701249045072b-10-01-2017.png&hash=ae6ac275925d8c8cf4921d2668e59a9655353048)
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F9429c6670d3609d74fdaae3582198c8d-09-01-2017.png&hash=050bccb58289a5042f6202f3019fb9d23d0bc1f0)
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F483ab541e337600f6d61289ac019e293-08-01-2017.png&hash=118162be9b7884e533bd5dce9f7565dd16a94ec8)
-
There's definitely congestion of some sort on the TTB network.
http://www.thinkbroadband.com/speedtest/results.html?id=148408595415694581972
Also gaming feels noticeably less responsive at peak periods.
The congestion isn't hard to spot on the BQM.
That's not a great speed test result, but thanks for posting. I'll pass the info onto Pulse8.
My BQM graph is looking pretty spikey as well. particularly overnight. Not sure what was going on there.
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2Fd13adc5c76924bbe10811fb70daadb6a-13-01-2017.png&hash=27c87d85b38a37bfccbdcbffd3a5df8e26a79bd2)
and F8lure graph for comparison
http://f8luresig.mouselike.org/sig.png?/57780.png
-
Hi
That graph reminds me of a similar one I saw using the ZyXEL 8924, regular intervals there were spikes.
Regards
Phil
-
Hi
That graph reminds me of a similar one I saw using the ZyXEL 8924, regular intervals there were spikes.
Regards
Phil
But this has only just happened since the last reboot following a power cut on Wednesday. I'm a dynamic IP address and have picked up a different subnet as a result. What's interesting is that the majority of speed testers were picking up my previous connection as Tiscali but are now saying TalkTalk.
I was about to post an earlier BQM graph, but have just realised the folly of deleting previous connections. There aren't any historical ones to post!
-
Sunday was a busy day for broadband it seems.
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F2eb1e46d32b8dc0ea04f440395ad2ba0-15-01-2017.png&hash=fa635a2553dfbab240f1e131277a155c56e0a159)
-
Hi
That doesn't necessarily show any problem with congestion. The green ping line is flat. Usually speed affecting congestion appears with the green following upwards as well, plus you get to see definite peak times in the graph rather than an all day of some pings with a slightly higher latency. See the examples at http://www.thinkbroadband.com/faq/sections/bqm.html.
Pings are usually treated as low priority so any gateway/router that is busy (but not necessarily congested) will treat pings with the lowest priority, including your own router.
Have you checked your data usage during a period you aren't using the connection yourself, for example overnight, just to make sure something isn't happening that is using bandwidth constantly, causing the congestion at your end?
My chart for Sunday using TalkTalk backhaul is attached. So this isn't typical of TalkTalk. Can you post a trace route from Thinkbroadband (make sure you remove your own IP address on the last step), as that might give some insight. http://www.thinkbroadband.com/tools/traceroute.html
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-thumb%2F5f0e7bb3f6fe6a0b75bccb8652b3b3ca-15-01-2017.png&hash=a78ad3f02464eec7223e37102788ea69bba87347)
Regards
Phil
-
Hi Phil,
I realise your response may have been aimed at Ignitionnet, however I thought it worthwhile adding my stats in here as well.
My BQM graph at the moment still looks spikey overnight and no idea what might be making calls on the internet during this time:
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F1047852e771452ab53ddea78de5f6fa1-18-01-2017.png&hash=2cbdf84d0234d54a3200cca44c2bfa566382dcf3)
My tracert to www.thinkbroadband.com is as follows:
# Host Sent Recvd Best Avg Worst
1 thinkbroadband-gw2.core-rs2.thdo.ncuk.net 15 15 0ms 2ms 3ms
2 po4-31.bdr-rt3.thdo.ncuk.net 15 15 0ms 10ms 157ms
3 ge-11-0-0-scr010.the.as13285.net 15 15 0ms 2ms 14ms
4 host-78-144-14-7.as13285.net 15 15 0ms 2ms 16ms
5 host-78-144-9-16.as13285.net 15 15 5ms 6ms 6ms
6 host-78-151-229-213.as13285.net 15 15 5ms 6ms 16ms
7 host-62-24-255-6.as13285.net 15 15 5ms 6ms 6ms
8 host-78-149-210-32.as13285.net 15 15 10ms 10ms 11ms
Thanks,
Jon
-
Hi
I wouldn't worry about the spikes, it is more red packet loss from the top and periods of green extending upwards that start to show some problem. The spikes can be caused by our own use of the internet connection as well.
My trace route is different even though it is still TalkTalk
1 thinkbroadband-gw2.core-rs2.thdo.ncuk.net 15 15 0ms 2ms 4ms
2 po4-31.bdr-rt3.thdo.ncuk.net 15 15 0ms 5ms 72ms
3 te2-3.cr05.hx2.bb.gxn.net 15 15 0ms 1ms 3ms
4 tbc.ar85.ts1.bb.daisyplc.net 15 15 0ms 1ms 1ms
5 tbc.ar85.ts1.bb.daisyplc.net 15 15 0ms 0ms 0ms
6 <me>.ip.static.uno.uk.net 15 15 9ms 10ms 11ms
Hopes 1 and 2 are the same, then I go across to gxn.net, who are GX Networks Ltd (used to be Pipex) and I think are taken over by TalkTalk, then from there it is all Daisy Plc, they are unconnected to TalkTalk, so presumably not much of TalkTalk is used, i.e. they aren't providing my internet connection, just a hand off to else where.
It looks like Pulse8 are just reselling a TalkTalk product through and through, so they aren't involved at all apart from raising the invoices, I might be wrong but that is what it appears.
So the problem is down to TalkTalk it seems, or a problem with the exchange and particular backhaul being used being congested, but it doesn't appear like congestion which is very much around peak times, and you are seeing slow single thread downloads 24/7, that's more like a deliberate restriction or throttle.
Can you change supplier?
Regards
Phil
-
Hi Phil,
Thanks for your feedback.
The only red downward spikes I see are when I am doing speed tests and putting maximum load on the connection. The green line is just that: a straight line at 12ms.
I also think the fact that I have tried both a high priority Business connection and lower priority Home connection without noticeable differences without suggestion suggests that this isn't straight forward contention.
At the moment, I'm more inclined to think that it is a TTB peering related issue. See below for something I posted on the TBB TT forum.
However if I test using TestMy.Net, a site that prides itself as a single thread test site, I get full speeds.
Thu Jan 12 2017 @ 12:00:37 pm up 6.1 MB 7.93 Mbps 992 kB/s
Thu Jan 12 2017 @ 12:00:19 pm down 200 MB 36.4 Mbps 4.55 MB/s
Thu Jan 12 2017 @ 11:00:34 am up 6.2 MB 7.7 Mbps 963 kB/s
Thu Jan 12 2017 @ 11:00:18 am down 200 MB 36.03 Mbps 4.5 MB/s
Thu Jan 12 2017 @ 10:00:28 am up 6.3 MB 7.78 Mbps 973 kB/s
Thu Jan 12 2017 @ 10:00:11 am down 200 MB 37.15 Mbps 4.64 MB/s
Thu Jan 12 2017 @ 09:00:28 am up 6.3 MB 7.85 Mbps 981 kB/s
Thu Jan 12 2017 @ 09:00:10 am down 200 MB 35.79 Mbps 4.47 MB/s
Thu Jan 12 2017 @ 07:59:40 am up 6.3 MB 8.57 Mbps 1.07 MB/s
Thu Jan 12 2017 @ 07:59:28 am down 43.1 MB 35.69 Mbps 4.46 MB/s
I have found a Dutch download test file site at:
http://test.unet.nl/1000mb.bin
that downloaded at 35.7 Mbps, which is in line with the TestMy.Net tests.
So results are variable depending on where you test.
Feels more like a peering issue to me.
or as MrS (aka Andrew responded)
Peering or path through providers network to the peers
Pulse8 are a TTB reseller pure and simple, but provide the front line technical support and billing facilities.
Changing suppliers is an option, but as I'm on TTB MPF, it's not a cheap business going elsewhere. Most of the ISPs I would choose to go to are quoting £75 upwards migration costs and none of them are willing to commit to saying they can improve my lot.
Bit of a gamble really.
Thanks,
Jon
-
Hi
Yes would appear to be some sort issue on a certain part of their network, seems that if you leave the UK it's okay, which probably switches you to a different route bypassing the problem.
Hopefully they can resolve it. Given the size of Thinkbroadband and the number of readers/publications that use it's ranking data, that TalkTalk would ensure speed tests to that site were running 100%. I would think if all TalkTalk customers experienced such slow single threaded speeds Mr Saffron would have commented before now as it would be clear from their speedtest data, perhaps he could check all Pulse8 customers?
Regards
Phil
-
Hi
Yes would appear to be some sort issue on a certain part of their network, seems that if you leave the UK it's okay, which probably switches you to a different route bypassing the problem.
Hopefully they can resolve it. Given the size of Thinkbroadband and the number of readers/publications that use it's ranking data, that TalkTalk would ensure speed tests to that site were running 100%. I would think if all TalkTalk customers experienced such slow single threaded speeds Mr Saffron would have commented before now as it would be clear from their speedtest data, perhaps he could check all Pulse8 customers?
Regards
Phil
Hi Phil,
It's probably more subtle than that as the TestMy.Net test server is in London, but it definitely seems to related to routing in some way.
I've got a couple of threads running on TBB and it seems that nobody on a direct TT or TTB connection can get a truly maxed out single thread result. Probably worth seeing if I can get MrS to respond though.
The best I have seen are the Uno connections, but it is obvious they are routed differently.
Do you know who Daisy use for the backhaul?
I will be chasing Pulse8 for an update on their ticket on Friday, if I haven't heard back from them before then.
Thanks,
Jon
-
Hi
Daisy are the back haul I think, not sure where TalkTalk hand over to Daisy.
Regards
Phil
-
That doesn't necessarily show any problem with congestion. The green ping line is flat. Usually speed affecting congestion appears with the green following upwards as well, plus you get to see definite peak times in the graph rather than an all day of some pings with a slightly higher latency. See the examples at http://www.thinkbroadband.com/faq/sections/bqm.html.
The green line is minimum latency within a 60 second period, based on samples taken every second in that minute.
Pings are usually treated as low priority so any gateway/router that is busy (but not necessarily congested) will treat pings with the lowest priority, including your own router.
Actually no, on most routers as far as the TBB BQM goes they aren't. If the router is just forwarding them it'll get as far as looking at the destination IP address, place them into an expedited forwarding queue and send them on. The reason carrier grade routers may deprioritise pings that have a TTL that requires a response from them as in the case of a traceroute, or have a destination of an interface on them is because they leave their routing plane and go into their control plane, so are dealt with differently from traffic they are simply forwarding. Home routers that have these issues are the exception rather than the rule as they tend to use an SoC which has its own routing components that include responding to pings.
CPU on my own router is pretty low as a general rule. It's rated for nearly 1Gb WAN to LAN throughput and is using cut-through forwarding for packets so is running pretty quietly. Plus, the spikes in latency persist even when not in use.
I've verified the jitter by pinging a couple of different locations at peak times. I have no reason to think my own equipment is at fault, especially given all was fine with the exact same set up before Christmas and before TalkTalk Business changed my LNS to one in London.
Have you checked your data usage during a period you aren't using the connection yourself, for example overnight, just to make sure something isn't happening that is using bandwidth constantly, causing the congestion at your end?
Yes, I have indeed. My router shows usage per client, per hour, alongside aggregated statistics.
-
It is very easy to see if the router is to blame, and that is to see what happens when you use the connection off peak, as the router has no concept of peak and off peak times.
I am pretty sure ignition is knowledgeable enough to already rule out his equipment. :)
-
Those are famous last words and shouldn't be taken for granted.
I have some work to do on the home network to prepare for dual-WAN again anyway which will eliminate my own router's PPP termination from the equation. I'm going to have the Huawei terminate the PPP tunnel and handle NAT for me, feeding an RFC1918 address to my router, with my router as its DMZ. This will allow use of full acceleration on the router as all IP.
I am pretty sure it will also mean the Huawei responds to pings on the WAN itself, however I'm not 100% sure. It may forward to DMZ.
Something else that makes me point the blame away from my kit is that I see the higher latency on pings through my router, and I feel the loss and jitter while gaming, but only at peak times. I have run an iPerf test off-peak through my kit capped at 50% of the link bandwidth, along with using QoS to prioritise the gaming traffic and latency and jitter were both virtually unchanged from baseline.
Going by when it happens I find myself wondering if there are too many customers connected to my LNS.
-
I think TTB may have been doing something as a result of the ticket that Pulse8 raised.
http://www.thinkbroadband.com/speedtest/results.html?id=148491160417956285839
I'm guessing that this is only an interim stage, as this is pretty impressive of a 38 connection!!
I've just chased P8 for some feedback on what is happening, but it looks as though Adam's positivity rating may be on the rise.
-
.... and the feedback from Pulse8 is that they think the improvement is more than likely associated to a change in the back end networking as all Pulse8 customers are being regraded to improved TTB packages.
Whatever the underlying cause, the improvement is significant.
Now trying to establish why my line is giving me >70 Mbps single thread results, when I am only paying for a 38 Home line.
-
Hi
.... and the feedback from Pulse8 is that they think the improvement is more than likely associated to a change in the back end networking as all Pulse8 customers are being regraded to improved TTB packages.
Whatever the underlying cause, the improvement is significant.
Now trying to establish why my line is giving me >70 Mbps single thread results, when I am only paying for a 38 Home line.
Your line and package must be back to 80/20, either temporary to allow them to get test results or a mistake has seen you go back to the faster package, whether they are charging you or not for the faster package or if it remains on your account, time will tell.
Good to see they are improving things.
Regards
Phil
-
Hi
Your line and package must be back to 80/20, either temporary to allow them to get test results or a mistake has seen you go back to the faster package, whether they are charging you or not for the faster package or if it remains on your account, time will tell.
Good to see they are improving things.
Regards
Phil
I've had an interesting dialogue with them this afternoon. It appears that they hadn't kept their admin up to date, so I had been upgraded to their 76XL package inadvertently.
Much as the temptation was to keep quite, I have been 'Honest Jon' and told them. They made an error with my January invoice, so I worked out what must have happened and I've elected to stay on their 76XL package now its giving full speed.
I can also confirm, it was the move of the Pulse8 lines to the new kit that has resulted in the speed difference.
I have seen the TTB response back to Pulse8 and I'm not impressed with what I saw.
There's no point in moving anywhere at the moment if the line continues to give full speed, but I will be looking for a bolt hole should the line degrade again.
I'm intrigued how A&A and Uno manage their TTB connections, as Pulse8 don't appear to get much leverage from TTB unless the line speed measured by the TTB speed test (multi thread and on their network) falls below the threshold level (~63 Mbps on my line).
I can't help thinking that the more technical ISPs (A&A, Uno, Zen, IDNet, etc) are capable of delivering a better service, albeit for a few more ££s.
Appreciate all your help.
Thanks,
Jon
-
The routing has changed, as might be expected.
# Host Sent Recvd Best Avg Worst
1 thinkbroadband-gw2.core-rs2.thdo.ncuk.net 15 15 1ms 10ms 120ms
2 po4-31.bdr-rt3.thdo.ncuk.net 15 15 0ms 1ms 15ms
3 ge-11-0-0-scr010.the.as13285.net 15 15 0ms 3ms 36ms
4 host-78-144-14-1.as13285.net 15 15 0ms 6ms 51ms
5 host-78-144-9-20.as13285.net 15 15 5ms 7ms 25ms
6 host-78-151-229-213.as13285.net 15 15 6ms 7ms 11ms
7 81.1.113.12 15 15 7ms 8ms 14ms
8 xxx-xxx-xxx-xxx.dynamic.dsl.as9105.com 15 15 10ms 10ms 11ms
compared to the old route:
My tracert to www.thinkbroadband.com is as follows:
# Host Sent Recvd Best Avg Worst
1 thinkbroadband-gw2.core-rs2.thdo.ncuk.net 15 15 0ms 2ms 3ms
2 po4-31.bdr-rt3.thdo.ncuk.net 15 15 0ms 10ms 157ms
3 ge-11-0-0-scr010.the.as13285.net 15 15 0ms 2ms 14ms
4 host-78-144-14-7.as13285.net 15 15 0ms 2ms 16ms
5 host-78-144-9-16.as13285.net 15 15 5ms 6ms 6ms
6 host-78-151-229-213.as13285.net 15 15 5ms 6ms 16ms
7 host-62-24-255-6.as13285.net 15 15 5ms 6ms 6ms
8 host-78-149-210-32.as13285.net 15 15 10ms 10ms 11ms
-
Hi
I'm intrigued how A&A and Uno manage their TTB connections, as Pulse8 don't appear to get much leverage from TTB unless the line speed measured by the TTB speed test (multi thread and on their network) falls below the threshold level (~63 Mbps on my line).
I'm not sure what the difference is exactly, but I assume Pulse8 are a reseller of TalkTalk products, it's essentially the same product just invoiced by a different company, whereas Uno and A&A purchase back-haul from TalkTalk just to get the customer into their own networks.
Good to hear you have it sorted out now.
Regards
Phil
-
I guess my issues might be more local in that case.
http://www.thinkbroadband.com/speedtest/results.html?id=148511885606282534130
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F98aba34acf9bcd5c978852179f7db4f3-22-01-2017.png&hash=0ee922de0d23eb10993f4ff94f92fe8a0bdc7efa)
Nice bit of congestion there, presumably due to a loss of a link, too.
-
Hi
I'm not sure what the difference is exactly, but I assume Pulse8 are a reseller of TalkTalk products, it's essentially the same product just invoiced by a different company, whereas Uno and A&A purchase back-haul from TalkTalk just to get the customer into their own networks.
Good to hear you have it sorted out now.
Regards
Phil
Pulse8 are a direct reseller of TTB Wholesale and do very little more than provide first line support and invoicing services. On the positive they do 30 day contracts on all products.
As your traceroute goes through the Daisy network, are Uno buying their backhaul from Daisy rather than directly off TTB?
Do you know if A&A purchase their back haul directly from TTB?
-
I guess my issues might be more local in that case.
http://www.thinkbroadband.com/speedtest/results.html?id=148511885606282534130
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-large%2F98aba34acf9bcd5c978852179f7db4f3-22-01-2017.png&hash=0ee922de0d23eb10993f4ff94f92fe8a0bdc7efa)
Nice bit of congestion there, presumably due to a loss of a link, too.
what was going on with that doubled base latency and large packet loss?
-
Hi
Pulse8 are a direct reseller of TTB Wholesale and do very little more than provide first line support and invoicing services. On the positive they do 30 day contracts on all products.
As your traceroute goes through the Daisy network, are Uno buying their backhaul from Daisy rather than directly off TTB?
Do you know if A&A purchase their back haul directly from TTB?
Yes they must be using Daisy for their backhaul. My Gateway is a TalkTalk gateway, after that it's onto the Daisy Network, so I suspect they purchase the FTTC connectivity from TalkTalk just to get their customers onto their own network. I don't know how A&A work, although I would think they do the same as Uno and are just buying connectivity from TalkTalk to get customers onto their own network.
Taken from https://www.talktalkbusiness.co.uk/partners/products/data/wholesale-connectivity/
Wholesale Connectivity provides a prioritised connection from the end-customer site through the TalkTalk Network and onto your network via an L2TP Interconnect. As a Partner, you can then add value to your customer by providing a range of internet, MPLS IPVPN or Voice services, all with great margin potential and exciting revenue opportunities.
So it will be different kettle of fish than Pulse8 just reselling a standard TalkTalk package.
Regards
Phil
-
Hi Philip.
Thanks for the additional information.
I agree that Pulse8's offering is significantly different to the way A&A and Uno TTB based product offerings. I'm a bit older and a lot wiser (or at least more knowledgeable!) than I was before I started this thread.
Cheers,
Jon
-
Not looking great now.
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.thinkbroadband.com%2Fping%2Fshare-thumb%2Fbf4d3decd3249ba33faf78e230944fd6-23-01-2017.png&hash=eb74ee4b58915ec4c9e6a25075b807f161d7c4f9)
Back on a 38 Product, even though I though i had agreed with Pulse8 to stay on 76XL.
I'm a bit confused as I was managing to access my work VPN and email whilst all that redness was going on.
Not feeling the love at the moment. >:(
-
why is the base latency so high? you interleaved?
-
Hi
Do you have a static IP address? Just wondering if your IP has changed and that chart is for someone else now who has picked up your IP?
Also sometimes a router can decide the Pings are some sort of attack and keep blocking them, this can become patchy if it's borderline on triggering.
Regards
Phil
-
Do you have a static IP address? Just wondering if your IP has changed and that chart is for someone else now who has picked up your IP?
Hi Phil,
Feeling pretty stupid now.
No I am on a dynamic IP address. Every other time my IP address changes, I have been presented with a wall of redness and because it was recording something I thought it had jumped back onto the same IP address with the product regrade.
I was somewhat surprised at the time, With the benefit of hindsight, it just feels naive! :-[
Ping traces now reset on f8lure and TBB BQM.
Request in with Pulse8 to move to a fixed IP address, once their regrade order (back to 76XL) is processed by TTB.
Bit frustrated with the apparent disconnect between Pulse8 and TTB at the moment, so hoping that will improve.
If not, I may be looking at other options.
Thanks,
Jon
-
I have refrained from playing back a blow by blow account of what is going on.
Pulse8 are progressing the issue with TTB, but only on the basis of my TTB speed test results having dropped by 10Mbps. TTB are really reluctant to take the TBB speedtest results into consideration.
Progress is really slow though and my thoughts are moving more in the direction of migrating away. I'm going to rule out A&A on the grounds of cost but Aquiss, IDNet, Uno and Zen are all under consideration.
Coming off TTB MPF onto BTW isn't cheap, but Zen and Aquiss offer the most cost effective route back.
Zen will cost £30 migration fee, but provide a ZyXel 1312 for a delivery charge. I'm hardly likely to use in in place of my VMG-8924, but possibly useful to have a back up router to jump past one of the hurdles when trying to log faults. Zen have confirmed that I would be on their LLU network rather than on BTW.
Aquiss have a three months fibre free offer on until the end of the month, which would cover the migration costs.
IDNet and Uno are not cheap options to migrate to just to prove a point so see them as less preferable from a cost perspective, however both Zen and Aquiss have had some history of single thread slow downs in the not too distant past, so they may be better long term bets.
I would welcome your thoughts on the above choices and indeed whether you think I would be foolish to jump from Pulse8 for the sake of a single thread speed that is roughly half what might be expected (see http://www.thinkbroadband.com/speedtest/results.html?id=148710995603882520677 ) when all my line parameters look good.
Any TTB speed test results or BQM graphs for fibre connections on any of the above would be much appreciated.
-
Turns out Uno are a non-option as they can't offer me a fibre to fibre migration path off TTB. And then there were three. ;)
Lots of proactive pre-migration support from Jon Green from Zen over on TBB, so they are probably in the hot seat at the moment.
My email chain with Martin Pitt at Aquiss dried up after an initial response so not as good a first impression, but will call them and IDNet over the next few days.
-
The guy on the Uno sales desk suggested that changing the MTU setting can improve single thread performance and early investigation suggested that may well be the case. My single thread speeds on the TTB tester improved; the slowdown I had seen on the TTB tester since being on a fixed IP disappeared.
However as I did more testing, the MTU settings that worked became more and more random and I started thinking it wasn't the MTU at all, but dropping the PPP session and re-enabling it that was critical.
Sure enough I've reset my MTU to 1492 and a couple of session drops later and my connection is showing full speed again.
Not sure why there should be good sessions and not so good sessions, but that appears to be the case.
I am assuming that because I can get good sessions that the issue isn't an issue with my configuration and the issue must be either within the TTB network or possibly the OR bit of the connection between my master socket and the exchange.
P8 are struggling to get anywhere with TTB, so it still looks as migration may offer me the best chance of proving it is the TTB network at fault.
Happy to stand corrected on any of this, so your thoughts would be appreciated (if I haven't bored you to death with this thread).
Decision time beckons. :-\
-
Hi jaydub
I could be wrong so apologies in advance, but I always worked on the theory to break a session, you disconnect for minimum 20 minutes, then reconnect, which should give you a different session. If connecting back within 20 minutes, your still connecting to old session
Many thanks
John
-
Doesn't look to be the case with me, as I seem to float between gateways when resetting the MTU (which causes a PPP session drop and restart) or manually stopping and restarting the PPP session. Some of them stick a bit, but I never go back onto the same gateway for more than 2 or 3 times at a go.
I have been doing a traceroute with each session restart and the pattern is not gateway specific.
Overall, about 40% of the sessions over the last couple of days have been 'good ones', but these are spread around the 4 different gateways.
-
Hi
Have you seen https://aastatus.net/2358, seems like a similar issue, could be related.
Regards
Phil
-
Hi Phil,
Thanks for the spot.
Pulse8 have managed to sort out whatever the issue was with TTB. As of 1st March the good sessions went from <40% to 100%. I see the occasional not so good TBB result, but for the most part I am seeing this now:
http://www.thinkbroadband.com/speedtest/results.html?id=148896057188254699524
Adam at Pulse8 believes the fix has been applied as a result of their escalation of the issue that they raised with TTB and is not related to the A&A issue.
The fact that the A&A issue is related to their Telehouse interconnect and still isn't resolved would seem to confirm this.
Pulse8 have just dropped their FTTC prices, so will be staying with them for at least the near future as long as the service remains as it currently is.
Thanks,
Jon
-
Can I ask a general, ignorant question: "How does it know?" (like the thermos flask, "the greatest invention of the century")
How does it know to be slow? Increased latency, or packet loss for no reason?
-
Well there's a question and a half.
I have absolutely no idea what was causing the issue. TBB BQM graphs are clean with a steady 12ms ping response. No dripping blood either.
See this one from 18th Feb, when I was suffering from slow single thread speeds:
http://www.thinkbroadband.com/ping/monitors/share/672c0ede7230c8c5ba2bb7974125a9a9-18-02-2017.html
compared to yesterday's when I was getting full single thread speeds:
http://www.thinkbroadband.com/ping/monitors/share/672c0ede7230c8c5ba2bb7974125a9a9-08-03-2017.html
No obvious difference as far as I can see.