New to IDNet - A few questions

[JamesAllen]

Hi there guys,

After many months of procrastination (probably going back over a year lol), I finally decided to move to IDNET. The reputation and service offering seems brilliant. Always encouraging to see how active and friendly this forum is as well!
Very nice to finally know that I will not have any restrictions on my connection 24/7. I was with Plusnet, who while a great providor were just too limiting in primetime. I'd often hit my primetime limits even though I was on a decent pro package (Premiere Option 1)..

So here I am..

Moved from ADSL Max to ADSL2+ on 21CN.

Anyway, my connection was activated at 2am this morning (very impressed considering it could have taken up to 6pm tonight).

I took some advice from the forum and picked up a Netgear DG834G V4 router - managed to get a brand new one from Ebay. Plugged in the settings and all seems good:

Downstream Sync: 14363 kbps
Upstream Sync: 1159 kbps

Now, my first question. Am I right to think that my BT profile could increase over the next 10 days?

When I first tested the connection it appeared the same as ADSL Max which I assume was the old profile for my phone line? That changed within about 15 mins. The BT Speedtester was failing until about 10 mins ago where I got these results:

---
Download speedachieved during the test was - 11786 Kbps
For your connection, the acceptable range of speedsis 4000-21000 Kbps.
Additional Information:
Your DSL Connection Rate :14363 Kbps(DOWN-STREAM), 1159 Kbps(UP-STREAM)
IP Profile for your line is - 12000 Kbps
The throughput of Best Efforts (BE) classes achieved during the test is - 15.23:28.73:56.04 (SBE:NBE:PBE)
These figures represent the ratio while sententiously passing Sub BE, Normal BE and Priority BE marked traffic.

Upload speed achieved during the test was - 897 Kbps
Additional Information:
Upstream Rate IP profile on your line is - 1159 Kbps
---

So all looking very good so far.

Looking at the above, do you think there is much chance of my BT profile going up to 13000 or 14000 or does it look about right to be at 12000 with a sync of 14363 on my router?

My second question though is speed tests. Obviously BT are showing the right speeds, but I found that only one speed test I tried (speedtest.net) shows download speeds of 11.1mb/sec when connected to a London node. Trying other locations gives much lower speeds and other speed tests give results from 2mbs to 5.5mbs.

Is this what you guys see as well? I.E. Many speed tests do not have the best connections?

Gotta admit, I'm very impressed with latency. 16ms on ping to bbc.co.uk. 

Just for the record, here are my router stats:

ADSL Link Connection Speed     Line Attenuation Noise Margin

Downstream       14363 kbps 25.0 db 5.8 db

Upstream 1159 kbps 11.1 db 6.1 db

Any comments and insight are greatly appreciated.

I am psyched to have a new quality ISP and connection. Also looking forward to hanging out here - you all seem like a great group of guys.

Cheers,
James.

[cavillas]

It depends on the servers, contentiion and routing on what download speeds are like from some sites. The best one to use is the BT broadband speed tester as it gives a fairly accurate result from exchange to user.

[JamesAllen]

It depends on the servers, contentiion and routing on what download speeds are like from some sites. The best one to use is the BT broadband speed tester as it gives a fairly accurate result from exchange to user.

Thanks Cavillas. It's encouraging to see the BT checker return the correct results. I think I was just expecting max throughput for some of the big speed testers from more than one particular location.

[Simon]

 James.  

[Ray]

  James

[Lance]

Your current profile is correct for your sync. However, if you resync when the noise is at its lowest you may be able to scrape a little higher and acheive a 13mb profile. I would use routerstats to monitor the noise margin for a few days to establish when noise is at its lowest - normally early morning.

http://www.idnetters.co.uk/forums/index.php?topic=1904.msg31527#msg31527

[JamesAllen]

Your current profile is correct for your sync. However, if you resync when the noise is at its lowest you may be able to scrape a little higher and acheive a 13mb profile. I would use routerstats to monitor the noise margin for a few days to establish when noise is at its lowest - normally early morning.

http://www.idnetters.co.uk/forums/index.php?topic=1904.msg31527#msg31527

Aha, awesome - that's perfect. Thanks Lance.

Yeah looking at that list my profile is spot on for my current sync rate. Excellent.

Resyncing at a particularly low noise point would just be a case of powering off my router and powering it back on again?

I'll look into routerstats - great to have those kind of stats available.

Thanks for your help.

[Steve]

  James

[Rik]

Resyncing at a particularly low noise point would just be a case of powering off my router and powering it back on again?

It depends on the router, James, some have software resets available. It's a while since I looked at a Netgear, but I seem to recall they do, off the connection information page. Bear in mind that a morning resync may get you extra speed, but it may also be accompanied by extra instability, so you could end up worse off.

Welcome to the forum.

[JamesAllen]

It depends on the router, James, some have software resets available. It's a while since I looked at a Netgear, but I seem to recall they do, off the connection information page. Bear in mind that a morning resync may get you extra speed, but it may also be accompanied by extra instability, so you could end up worse off.

Welcome to the forum.

Thanks Rik - much appreciated.

Ah I was wondering about that as while I could get a profile boost, if noise levels then rose, like you say it could worsen the connection.

Things are looking good so far so I'm not too worried.

My only slight concern is why only one speed tester from one location (London) is showing the proper downstream.. All the others show 2mb - 5.5mb or so. There is no reason why IDNet would have slow links to certain locations is it? Stupid question I know but I didnt do too many speed tests from other locations when on Plusnet so I'm not sure what I should be seeing.

[Rik]

Speed tests are, in general, unreliable and the only one which counts (if you can get it to work) is the BT one. I find that the speedtest.net Maidenhead server will give a very close approximation to the BT test, but if the servers are busy, all bets are off.

[armadillo]

I find that the BT speedtester usually gives me rather slower speeds than reported by speedtest.net, sometimes a lot slower. I find that the best true guide to what my line is achieving is to use the graph and speed meter in the networx software http://www.softperfect.com/products/networx/ that I learned about from the friendly folk on this forum. The way I use it is to download several largish files simultaneously from several different servers and then measure the speeds and check the graph. The speed is then pretty consistently the same as my IP Profile even though the BT speed tester rarely fully achieves the profile throughput.

It is also interesting to see the effect of congestion by looking at the graph. When the graph has a nice level top to it, there is no congestion and the line remains fully loaded throughout the downloads. When the graph has a "combed" appearance, there is  congestion between me and the server and the download is stop-start. If I do a BT speed test followed by a speedtest.net speed test, the graph often shows combing on the bt one but the speedtest.net one is level. Since all data has to pass through idnet and the BT exchange, that suggests that the congestion is on the BT server or along the route from the BT server to idnet. The BT speed test tends to be closest to the speedtest.net one and the download of real-world files in the wee small hours like 2am.

Multiple simultaneous downloads usually give a nice flat top to the graph unless I am downloading at a very congested time like 7pm.

[JamesAllen]

I find that the BT speedtester usually gives me rather slower speeds than reported by speedtest.net, sometimes a lot slower. I find that the best true guide to what my line is achieving is to use the graph and speed meter in the networx software http://www.softperfect.com/products/networx/ that I learned about from the friendly folk on this forum. The way I use it is to download several largish files simultaneously from several different servers and then measure the speeds and check the graph. The speed is then pretty consistently the same as my IP Profile even though the BT speed tester rarely fully achieves the profile throughput.

It is also interesting to see the effect of congestion by looking at the graph. When the graph has a nice level top to it, there is no congestion and the line remains fully loaded throughout the downloads. When the graph has a "combed" appearance, there is  congestion between me and the server and the download is stop-start. If I do a BT speed test followed by a speedtest.net speed test, the graph often shows combing on the bt one but the speedtest.net one is level. Since all data has to pass through idnet and the BT exchange, that suggests that the congestion is on the BT server or along the route from the BT server to idnet. The BT speed test tends to be closest to the speedtest.net one and the download of real-world files in the wee small hours like 2am.

Multiple simultaneous downloads usually give a nice flat top to the graph unless I am downloading at a very congested time like 7pm.

Awesome tip Armadillo - like it.

It's that kind of insight that I'm looking for when testing my connection as it can nicely answer the question of where the speed problems lie or the current state of the BT exchange.

More info the better in my opinion.

[Rik]

Always.

[armadillo]

You're welcome James. I've had so much help and info from this forum, it's nice to give something back.

[Rik]

 

[esh]

From my experience, I have found speedtest.net to be the most accurate. I measure raw throughput in 1 minute averages going through my server and compare it to the numbers output by speedtest, and they agree to within 0.02 Mbit/sec.

TBB on the other hand, I have found routinely hopeless. Either it's the server that gets picked for me or it's just overwhelmed.

[JamesAllen]

From my experience, I have found speedtest.net to be the most accurate. I measure raw throughput in 1 minute averages going through my server and compare it to the numbers output by speedtest, and they agree to within 0.02 Mbit/sec.

TBB on the other hand, I have found routinely hopeless. Either it's the server that gets picked for me or it's just overwhelmed.

Yeah I must admit I find speedtest.net to be the best. The server it detects to do the tests is generally the right one. When I try others they often don't show my max throughput.

TBB does seem rubbish.. 2mbs.. yeah right.

[armadillo]

I agree that speedtest.net is usually the fastest and TBB is the slowest. I think that TBB is the only one that uses a single thread for the download. The others multithread and that can make a big difference. Speedchecker.org is another good one.

I do not know what OS you are using. If you are on XP and have not adjusted your RWIN setting, multithreading can make a huge difference. Tweaking RWIN does not help with Vista or 7 though. I doubled my TBB throughput after tweaking RWIN on my XP system. But it is still the slowest!

[Rik]

There's speed tests and then there's the real world, eh?