capacity issues - continued
Sorry for the delay in responding guys.
Sizeable post on the way.
[EDIT]
First up, we really apologise for not replying to this thread for so long. The short answer to why is that there hasn’t been a lot to say. The longer is a bit more complicated.
We think there’s more than one issue affecting people, some of which we have more details on than others.
The main issue is fairly easy to see and especially impacts customers with the fastest fibre speeds. At peak times single threaded downloads slow down while multi-threaded downloads stay at pretty much full speed. Our efforts have been concentrated on identifying the cause of this problem. Unfortunately so far we’ve not been able to pinpoint the cause of the problem. We’ve gone through every piece of kit on our network with a fine tooth comb, we’ve done thousands of tests and monitored the interfaces and traffic levels, CPU load and memory usage and not found a single thing that would be the cause. We’ve repeated the tests with our various suppliers and they too have failed to find anything amiss that would cause the problem. We’ve tried correlating customers with when they are seeing a slowdown and when they aren’t and no pattern emerges. If it was something simple like a congested interface, a device running beyond its design or something slowing traffic down when it shouldn’t we would have long since found it.
We can replicate the results on our 80Mbps FTTC line here in the office, we’ve also used a 300Mbps line and seen the problem there too. We know the problem is real and it has the focus of our team to resolve.
We’ve managed to eliminate some things that could be the cause and there are many things which are very unlikely to be the cause but nothing we’ve been able to say “it’s definitely this”. In our testing we can see times when we get full speed all of the time, we also see times where it slows down just at peak and we also see times when it slows down gradually over the day. The best way of describing it is to imagine there are multiple routes from you the customer to the Internet. Each route will traverse different pieces of kit and different fibres in both our network and those of our suppliers. Any customer could take one of hundreds, possibly thousands of different routes. It’s very similar to driving, if I’m going from Sheffield to London I can drive down the M1, catch the train, take the A roads or even drive up to Leeds and go across the M62 and down the M6.
One thing we do know is that when you look at the Thinkbroadband latency graphs you don’t notice any change in latency or packet loss. If you see these changes in these the problem is something different.
Back in the thread some people have noted seeing a couple of people switch ISPs and see the problem go away. We’ve spoken to our colleagues over at BTR (BT Retail) to try and see if we can understand why, we know their network is different to our network (only a part of the path between the customer and ISP is common to both) so we think this helps to eliminate certain parts of the route but doesn’t actually pinpoint the cause. Same with someone moving to other ISPs, again part of the route will be common but there will be a point where the routes diverge.
We’re trying to set up some testing which we think will eliminate virtually everything except one particular point, this testing is taking time to setup but we think it will be worth it as it tell us where abouts the problem is.
We know there are other issues too, the last few days some Thinkbroadband graphs have been showing some packet loss at peak (e.g. marmalizer, OshB5 and Strat) we’ve identified the cause of this and should have it fixed soon.
To those people that are posting speedtest and other results we really really appreciate it. This thread is read by many people in the business even if they haven’t replied, they all know that it isn’t a problem isolated to a few people as we’ve demonstrated the problem to them on our office line for them to see.
Sizeable post on the way.
[EDIT]
First up, we really apologise for not replying to this thread for so long. The short answer to why is that there hasn’t been a lot to say. The longer is a bit more complicated.
We think there’s more than one issue affecting people, some of which we have more details on than others.
The main issue is fairly easy to see and especially impacts customers with the fastest fibre speeds. At peak times single threaded downloads slow down while multi-threaded downloads stay at pretty much full speed. Our efforts have been concentrated on identifying the cause of this problem. Unfortunately so far we’ve not been able to pinpoint the cause of the problem. We’ve gone through every piece of kit on our network with a fine tooth comb, we’ve done thousands of tests and monitored the interfaces and traffic levels, CPU load and memory usage and not found a single thing that would be the cause. We’ve repeated the tests with our various suppliers and they too have failed to find anything amiss that would cause the problem. We’ve tried correlating customers with when they are seeing a slowdown and when they aren’t and no pattern emerges. If it was something simple like a congested interface, a device running beyond its design or something slowing traffic down when it shouldn’t we would have long since found it.
We can replicate the results on our 80Mbps FTTC line here in the office, we’ve also used a 300Mbps line and seen the problem there too. We know the problem is real and it has the focus of our team to resolve.
We’ve managed to eliminate some things that could be the cause and there are many things which are very unlikely to be the cause but nothing we’ve been able to say “it’s definitely this”. In our testing we can see times when we get full speed all of the time, we also see times where it slows down just at peak and we also see times when it slows down gradually over the day. The best way of describing it is to imagine there are multiple routes from you the customer to the Internet. Each route will traverse different pieces of kit and different fibres in both our network and those of our suppliers. Any customer could take one of hundreds, possibly thousands of different routes. It’s very similar to driving, if I’m going from Sheffield to London I can drive down the M1, catch the train, take the A roads or even drive up to Leeds and go across the M62 and down the M6.
One thing we do know is that when you look at the Thinkbroadband latency graphs you don’t notice any change in latency or packet loss. If you see these changes in these the problem is something different.
Back in the thread some people have noted seeing a couple of people switch ISPs and see the problem go away. We’ve spoken to our colleagues over at BTR (BT Retail) to try and see if we can understand why, we know their network is different to our network (only a part of the path between the customer and ISP is common to both) so we think this helps to eliminate certain parts of the route but doesn’t actually pinpoint the cause. Same with someone moving to other ISPs, again part of the route will be common but there will be a point where the routes diverge.
We’re trying to set up some testing which we think will eliminate virtually everything except one particular point, this testing is taking time to setup but we think it will be worth it as it tell us where abouts the problem is.
We know there are other issues too, the last few days some Thinkbroadband graphs have been showing some packet loss at peak (e.g. marmalizer, OshB5 and Strat) we’ve identified the cause of this and should have it fixed soon.
To those people that are posting speedtest and other results we really really appreciate it. This thread is read by many people in the business even if they haven’t replied, they all know that it isn’t a problem isolated to a few people as we’ve demonstrated the problem to them on our office line for them to see.
Thanks for the response Mathew 
hopefully it will alleviate some peoples worry's for lack of input from PN 
hopefully it will alleviate some peoples worry's for lack of input from PN
@AndyH
Now do you believe the issues are real?
Quote from: Matthew Sorry for the delay in responding guys.
Sizeable post on the way.
[EDIT]
First up, we really apologise for not replying to this thread for so long. The short answer to why is that there hasn’t been a lot to say. The longer is a bit more complicated.
We think there’s more than one issue affecting people, some of which we have more details on than others.
<snip>
Now do you believe the issues are real?
Here is the before and after I moved gateway.
Quote from: Matthew One thing we do know is that when you look at the Thinkbroadband latency graphs you don’t notice any change in latency or packet loss. If you see these changes in these the problem is something different.
I have really truly appalling browsing latency today and buffering with streaming as well (which is unusual,) yet TBB latency is perfect and nothing unusual on a speed test either (although it's an 80/20 connection capped at 40Mbps by PlusNet, so that might be why; and the lower-than-expected upstream speed does continue).
I've been too busy to sort out migrating to another ISP before the end of the 30-day price rise notice window, unfortunately, so I'm relying on PlusNet to fix the problem now Matthew.
I sincerely hope things get better...
