So I'm a 350mbit business user in Frome. I was previously having a pretty decent service at my last premises (in town centre) and was happy. Ive now moved to a new location further out of town and my network performance has been dire.
Initially there was a lot of physical issues and over 3-4 visits the engineers replaced a load of infrastructure and got their various channels SNR values into their spec. They saw that speed seemed low but said they couldnt do anything about it as outside of the scope of what they can do.
When you test via speedtest.net the connection is apparently 300-350mbit every time you test. However real-world performance (http, ftp, rsync, git...) is nowhere near that. Fast.com reports an average of 140mbit speed which seems pretty close to what we're seeing as real-world performance.
Now I asked customer service they've temporarily upped the service to 500 mbit to see if it makes a difference. By some strange co-incidence speedtest.net reports 500-600mbit each time you test and fast.com reports around 230mbit consistently.
So real-world performance seems to be around the magic 50% mark of advertised speeds which VM deem "acceptible" in both cases. i.e. the cable is quite capable of delivering the speed Im paying for, it appears that the network reduces bandwidth to the minimum that can be delivered.
I thought that seems a bit weird so I experimented a bit further. What I found was that despite fast.com *always* reporting around 230mbit now, if I run a speedtest.net test and then initiate a fast.com test near to its completion, suddenly fast.com reports 600mbit.The plot thickens.
I then compiled a copy of netspeed (https://github.com/nedko/netspeed) and tried running upload only tests at the same time as doing a fast.com test and bingo, throughput increases to 600mbit whist it's running and decreases after it stops.
So this seems pretty clear to me that there's a load of traffic prioritization by VM going on at the network level. I tried explaining this to a member of the fault team via chat who claimed that it's not a fault (!)
What can one do to get VM to actually investigate this and try and solve it? There was never this problem at the old premesis and one can only conclude that in the best case theres a mis-configuration somewhere and in the worst case, deliberate traffic shaping.
That's interesting, there was nothing I saw in the forums saying it was residential not business? I'll have a go at replicating the same with a residential router / account on the same line as well then.
I've already been down the road of discussing with several agents via the business support links. They seem as keen as possible to not investigate at all ; hence posting here.