Are BT Wholesale prioritising their Speed Test traffic?

I have a Zen VDSL connection. It syncs at 80/20 and through the day it runs great at 60+Mb down, 20Mb up.

In the evenings, it ALWAYS drastically slows down, it can be anything between 2Mb and 60Mb, typically hovering around 10Mb and at times it is even slower than my 7.5km long backup ADSL line at ~2Mb.

Tonight I did several speed tests on Zens own tester, several on Think Broadband, several on Ookla and several on the BT Wholesale checker. Every result from Zen, TB or Ookla was less than 10Mb down. Every test from BT Wholesale checker was 50+Mb. Neither Netflix or Amazon will stream in HD but BT Wholesale checker says I am getting 50Mb+

I smell a rat! Its like BTW are rigging their speed tests to look better and mask problems.

The BT Wholesale test on some providers will miss part of the network, so if giving good speeds and others are not can point to where a problem would be.

Also worth verifying the throughput during the test to check the tester is showing sensible results rather than just a very short burst of speed

Thanks Andrew.
I have DSLstats monitoring the connection and the traffic/minute graph peaks at 20Mb down through all the tests yet on every one of the BT Wholesale speed tests, it was stating 50Mb+

Need to use a per second monitoring of the traffic or better to spot this sort of thing

Network tab in Window Task Manager is usually enough

I would show them you tbbqm obvious congestion on it imo if that isn't usage from around 4pm to 11:30pm , When i was on a Over contended SVLAN the BTW speedtester would show inaccurate results quite often always higher on the downstream and sometimes ridiculously low on the upstream , but when downloading files from a reliable source if you are seeing erratic throughput then it's congestion , likely within BTW's realm

And thinking about it Zen tech support should be able to see that packet loss and change in base latency at peak times , so should be able to raise with BTW, (once they have set up monitoring on your circuit)

Edited by tommy45 (Mon 11-Jan-16 00:57:58)

I would give very good odds on this being due to very different routing. The BT Wholesale speed testing sites will be integrated into the BTW network and not have to disappear across some peering link to get to the Internet. The BTW speed tester will have been specifically designed to test throughput within the BTW network, not that across peering connections to the Internet. So things like the backhaul connection from the exchanges, FTTC cabinet and so on, not general Internet performance.

Network topologies are very complex and what I suspect you are seeing is congestion on some peering like to the Internet. Just where that is in the link from you to those Internet test sites and who is doing the capacity planning (and paying for it), I've no idea, but your first port of call ought to be Zen if it's causing issues.

You can sometimes work out which hops on the network are causing the delay by running tracert during both good times and bad times and seeing what changes. However, tracert won't necessarily see all the hops (because of the way that traffic is piped across the BTW network) and many routers don't respond to the required protocols.

Edited by deleted (Mon 11-Jan-16 09:28:58)

The testers have two functions.

The BT test is really there to demonstrate that the line is performing correctly - yes there might be minimal congestion as it bypasses a lot of the network and is directed straight to BTs servers.. The other testers are to show what can be delivered at any time and will be subject to network congestion.

Similar situation here with my Vivaciti FTTC.
Until recently daytime congestion resulted in low TBB result amd similar BTW result but BTW tap3 test showed full >70 Mbps.
Last few weeks TBB showed low congested result and 1st stage BTW shows full 70Mbps.
At least before I could show supplier vastly different results from BTW stage 1 and tap3 tests.