Generally 2-3Mbps below BT Consumer it seems. What is unclear is the reason for it. The argument is with those who blame it on the Plusnet copy IP Profile system, which is just far too simplistic.

I'm pleased about the similarity you post between BT and EE . That veers towards what something/somebody "sort of authoritative" but I can't remember the source said a couple of years ago.

That BT Consumer broadband doesn't feed into MSILs from the BTW Cloud, but goes straight to BTW peering outwards. (Isn't that rather like BT Central Pluses worked? I think they may have been mentioned at the same time).

How to confirm or rubbish that I don't know. But what we do know is that EE must be working like that, as EE doesn't have its broadband traffic fed from BT Wholesale. (It is possible they do still have mail hosting under their own control of course). They sold the whole operation to them, so it appears to be White Label. Could BT Consumer also be such is the question that arises.

If this is the case, straight away we have a possible explanation of the higher speeds that BT achieve.

The thread arose of course from my empirical results of near-identical speed test results on my non-resync'ed line at similar times. Plus this afternoon after a re-sync figures comparable with historic Plusnet results I have.

For historic figures I opened up my thinkbroadband test results and looked for very close matches to today's results. Then I browsed through all my TTTS posts to find corresponding dates, and checked my syncs as shown in my sig for them.

It is quite clear that Plusnet is no slower on speedtests than AAISP. Normal usage feels better.

However, as is often the case when Plusnet speeds are mentioned, disruption and distraction ensue from a particular poster, thus preventing rational discussion of what is going on.

Hopefully we are back on track .

I'm not quite sure what you are saying with "there is a gap to the rest who are evenly spread between 0.8-2.0 Mb/s lower". By "the rest" are you referring to the rest of Plusnet lines, or the rest of the ISPs, and if the latter which is how I initially took it, are you saying they are below Plusnet?

In reply to a post by MrSaffron:

http://blog.thinkbroadband.com/2014/12/detailed-anal...

Has speed profiles from the main providers

I would think that sam knows data to be possibly more accurate as there probably will be a bigger number of samples , it corresponds with my own experience when with plusnet, unlike the results that suggest that throughput levels increase during peak times which is more than likely due to customers who are /have been unaffected by the peak time issues, running speed tests at those times, or some other occurrence i can still recall trying to conduct a speed test on tbb, and the speed tester failing to load due to insufficient bandwidth, or not completing on occasions when things where that bad

Edited by tommy45 (Sun 16-Aug-15 23:34:49)

In reply to a post by RobertoS:

Generally 2-3Mbps below BT Consumer it seems. What is unclear is the reason for it. The argument is with those who blame it on the Plusnet copy IP Profile system, which is just far too simplistic.

really when they need that secondary profile for their traffic shaping , Regardless of what bit of kit in the chain at plusnet is responsible for the lower throughput levels if they did not use a secondary ip profile as a way of capping throughput then it couldn't be deemed as responsible or in part responsible could it? to go from a max throughput of 72.?? 73 mbps to 74-75mbps by swapping isp higher if i was with sky (75-76mbps ) but they aren't on BT WBMC with it's IP profile are they they are BTOR GEA

Edited by tommy45 (Sun 16-Aug-15 23:46:24)

I mean 90% of the Plusnet lines were 0.8-2.0 Mb/s slower than the corresponding BT and EE lines.

I don't think you can conclude anything from your single result in isolation - e.g. you may have been in the 10%. I also don't think anything about BTw infrastructure offers an obvious explanation for Plusnet's slower speeds (although I think EE even though end-to-end managed uses WBMC dedicated, unlike Plusnet who use WBMC shared)

Just seen this, but need some shut-eye. Thanks for the extra info .

thanks for the feedback, good for the community.

So if I understand right, there is still visible BTw issues but nowhere near as severe as plusnet, and in addition various activities perform faster.

On sky for me is similar, web browsing is definitely snappier, speedtests also have always been perfect aside from that one night when tbb had an issue.

Interestingly as well ipv6 via a HE tunnel works better than native ipv6 did on plusnet, not had the same mtu issues, but this maybe down to the fact sky dont use pppoe so is natively 1500 bytes mtu.

Also on plusnet I had weird issues with pages occasionally not loading or hanging, the most frequent was the bbc website, those issues have gone since I migrated.

tommy I have to say whilst I agree your issues with plusnet were mostly valid.

The profile issue isnt one of them, plusnet set it below the BTw level, but the % difference is small enough to not really be relevant, plusnet have much bigger issues than the profile. In fact their main issue with the profile system is the fact it can get stuck rather than what it is set at.

I have QoS/policing on my own router which is set below my line rate, this is to manage bufferbloat, I consider the loss of max burst speed acceptable, but I have excempted samknows (and now also tbb speedtester) so I dont provide bad data to samknows testing.

Interestingly I compared my tbb to jchamier who also is a samknows tester and his spikes are noticeably worse. When I used QoS on my router on plusnet the spikes were reduced, if I do the same now on sky they dissapear completely. So plusnet's QoS I feel degrades the connection.

Edited by Chrysalis (Mon 17-Aug-15 05:41:50)

If 90% then should be visible from profile plots.

Samknows is good for profiling a line but sample sizes otherwise are very low when you consider 2000 boxes for six providers, adsl & fttc products and different line length profiles and location across uk

I think you need to get over the sample size thing.

Still waiting for you to publish data that is only from logged in users and who have plotted tests against peak and off peak. Even if its only 5 users its better.

It cant be hard given the small amount of users, it just seems something is motivating you to not do it.

Using anonymous tests with no tracking to individual lines is clearly not the way.

Edited by Chrysalis (Mon 17-Aug-15 07:57:10)

Well I offered to give unique tracking versions some time ago and people are so concerned that no-one came forward.

Love how you say something is motivating me not to do it, I'd love to be able to wave a red flag saying look an ISP has very poor speeds that contradict every other outlet on speeds

Only reluctance is spent two days so far and inclusive results so far and add the many times I've sat in an evening looking at peoples results and the pattern is say four tests from a user and 1 is bad three are good and even when people say its falling apart.

http://labs.thinkbroadband.com/speedtest/index.php?s...
Is a unique version that will do an auto detect of postcode (may or may not work) but will test people every five minutes (8 seconds download, 8 seconds upload so not too much capacity used)