10Mb broadband - ok to UK sites but US only 400Mb
Discussion
I just got updated to a 10Mb service with a popular ISP. When I was on their 4Mb service I pretty much got 4Mb consistantly. Now I get 10Mb to UK sites, but US sites have slowed right down to 400Mb. This seems pretty strange, so I wanted to check if anyone else has noticed this sort of thing before having a go at the ISP.
victormeldrew said:
I just got updated to a 10Mb service with a popular ISP. When I was on their 4Mb service I pretty much got 4Mb consistantly. Now I get 10Mb to UK sites, but US sites have slowed right down to 400Mb. This seems pretty strange, so I wanted to check if anyone else has noticed this sort of thing before having a go at the ISP.
How are you measuring the amount of traffic you can get through the link? I presume you're getting 400Kb/second from the US?
The download speed of a single TCP connection also has many limiting factors regardless of your link speed so it may be working perfectly - it's just you're too far away (and therefore have high latency) to utilise your new link speed.
D
I've tried a few different speed test sites. myspeed.visualware.com allows selection of UK and US servers for the tests, and thats where the difference comes from. Last week when I was on 4MB I got 3.9MBish from US and UK servers.
I just tried another site, where I get 9Mb UK, but only 2MB to Moscow, and to Yokohama just 148kbps (less than the upload speed to the same server!).
This seems to be common across all the speed test sites I am trying, and my perception of general browsing speed is that following the "upgrade" my web browser - and in particular googling - is slower than it was.
I just tried another site, where I get 9Mb UK, but only 2MB to Moscow, and to Yokohama just 148kbps (less than the upload speed to the same server!).
This seems to be common across all the speed test sites I am trying, and my perception of general browsing speed is that following the "upgrade" my web browser - and in particular googling - is slower than it was.
I would of thought it has something to do with the number of hops, or servers the traffic has to pass through. The more hops the more potential for bottlenecks. Have you tried traceroute (tracert), been a long time since i used it, but if i remember it tells you the latency for each hop/router.
Edited by scorp on Sunday 17th September 23:34
I think that's about par for the course.
Over high latency links, your download performance will be directly related to the latency; more latency, less speed.
The other issue is that there's a higher chance of packets being dropped, the more routers that are encountered on the way.
The only other possibility is that your ISP's transatlantic bandwidth is overutilised. Most ISP's route Asian traffic through the US, so anything outside of Europe will be slow if that's the case.
D
Over high latency links, your download performance will be directly related to the latency; more latency, less speed.
The other issue is that there's a higher chance of packets being dropped, the more routers that are encountered on the way.
The only other possibility is that your ISP's transatlantic bandwidth is overutilised. Most ISP's route Asian traffic through the US, so anything outside of Europe will be slow if that's the case.
D
[URL=www.speedtest.net]
[/URL]Seems a bit faster today, yet latency is less?
[/URL]Seems a bit faster today, yet latency is less? scorp said:
aldi said:
Latency and bandwidth are related, but quite loosley.
The best analogy is a bit of pipe, latency is how long, bandwidth is how wide.
Edited by scorp on Monday 18th September 22:05
I'll go with that, apart from when pouring water through it, the guy on the other end needs to tell you to stop when a container is full to allow him to empty it. The longer the pipe, the more difficult/time consuming the conversation is. There's also more material, so a larger chance of the pipe cracking and losing water.
aldi said:
Latency and bandwidth are related, but quite loosley. The results you got before you upgraded to 10 meg, how many times did you run the test?
Not that often to be fair. I was having a problem with a modem overheating, and performance degraded noticeable for a while before it lost connection altogether, so I ran the speed tests at http://myspeed.visualware.com whenever things started lagging. I don't recall ever having to switch to the London server on that test, yet reliably got 3.8Mb+ unless my modem was overheating. If it was, I powered it off for 5 mins and normal service was duly restored. I have a bracket to mount in vertically now and it is no longer causing problems. I will admit there is a possibility that I might just be noticing a phenomenon that has always been there, simply because something else has changed. And I get so p*ssed off with my users when they do that!
If the provider is delivering your connection over their own LLU network rather than running off BT then there is a fair chance that they are doing much cleverer things with cacheing and peering then BT do. Most LLU networks have excellent peering at LINX which allows them good access to UK sites and access at high speeds, they will be combined with caching often down to the regional or even local exchange level which will affect the speeds shown by testing sites like visualware. Where they make back the vast cost of building the network while maintaining low prices for a better service - an anomaly clearly - is by limiting their exposure to the higher cost of Transit bandwidth which needs to be purchased from major carriers and costs 100 times what bandwidth costs at LINX.
It is important to remember that they only offer you a 10Mbps connection to their networks or at its most convoluted to "the Internet" (which could mean anything) they don't sell you 10Mbps to anywhere. In addition they publish the contention rate at your local exchange, not the contention rate that they apply deeper into their network which is far far higher.
All of this comes into play before latency affects traffic and if you are getting something from Japan the rubbish quality of connections over the south china sea.
Basically in broadband as in many other things if they sell you something and it seems like too god a deal there is a reason, they will have cut corners, worded the marketing and contract carefully and will be ready to sell you phonelines, 'protect your family' security solutions and on-demand TV as soon as they think you are not looking. Chasing better quality connections is probably going to be like chasing your tail.
It is important to remember that they only offer you a 10Mbps connection to their networks or at its most convoluted to "the Internet" (which could mean anything) they don't sell you 10Mbps to anywhere. In addition they publish the contention rate at your local exchange, not the contention rate that they apply deeper into their network which is far far higher.
All of this comes into play before latency affects traffic and if you are getting something from Japan the rubbish quality of connections over the south china sea.
Basically in broadband as in many other things if they sell you something and it seems like too god a deal there is a reason, they will have cut corners, worded the marketing and contract carefully and will be ready to sell you phonelines, 'protect your family' security solutions and on-demand TV as soon as they think you are not looking. Chasing better quality connections is probably going to be like chasing your tail.
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff