Pages in this thread: 
Print Thread
Most likely congestion, Zen run one of our speed test appliances on their internal network (does not feed our main speed test database) so contact support to get a support number and try running that and comparing to the external tests.
If external and internal are both bad, then points towards a problem between the cabinet and the zen internal network at peak times.
I'm guessing your appliance within Zen is the one at http://speedtest.zen.co.uk.
I have the same issue as Kamil and aparat, and I get identical results whether using the TBB test hosted on your side, or on Zen's network. It also affects basically anything single-threaded regardless of origin - e.g. HTTP downloads in a web browser.
I don't believe cabinet congestion to be the issue, because that would surely impact multi threaded performance? I get absolutely flawless speeds if I can use a download manager or something like bittorrent. It's universally poor too, regardless of time of day (though I haven't yet tried it at 3am). I think Zen did tell me that Openreach didn't report any congestion issues?
Here's a more extreme example from my connection: http://www.thinkbroadband.com/speedtest/results.html...
Though occasionally it's fine: http://www.thinkbroadband.com/speedtest/results.html... (ignore the fact that it's IPv6, when it's poor it's poor with either protocol - and this is Zen's native IPv6, not a tunnel)
Right now it's all over the place on single threaded, but absolutely solid with multi threaded: http://www.thinkbroadband.com/speedtest/results.html...
I've noticed one instance where the BT wholesale speedtest gives me poor results, but doing the test3 while connected to the BTwholesale test network gives me great results.
The difference between the multi-thread and single thread download tests are interesting.
As far as I can tell/understand the ISP network sees no difference between the two.
A single test will have one data stream between your routers IP address and the test sites IP address. I assume a multi-thread test uses the same IP addresses but with different originating port addresses. The port addresses are part of TCP protocol while the ISP network is responsible for the IP part - TCP on IP.
I wonder if you have a firewall that's not great on individual threads.
I use pfSense on a beefy box (and have CODEL enabled to try to handle bufferbloat), so my router or lack of performance is not the issue. But I get exactly the same result even if I connect a computer directly to the Openreach modem, so my entire network can be ruled out.
My understanding of Zen's POP rollout is that they have kit in many BT exchanges and the backhauls from the exchange to Zen's core network are under Zen's control. Unfortunately I've not found a list of exchanges with POPs.
There's a map showing their PoPs if you click the link in their press release. I'm not on the "Zen network", I'm definitely connected via BT wholesale.
Edited by deleted (Sat 30-Jan-16 11:22:27)
) to be causing a problem so why the single thread speed does not max out the connection is very odd.