TBB speed test

I've just found the new TBB speed tester and I'm greatly impressed. Congratulations to those who developed it and thanks for making it available to us.

I feel all warm and fuzzy inside now

This is not just a one off update either, lots more tweaks and presentation of results is planned and support for wider platforms building on the Android app that I did a few updates to in 2012.

It's great, though I do miss seeing the graph which showed me how my results have changed over many months or years. Is that one of the tweaks which are to follow?

Very much so. The aim was to release and get database building up so I can look at the issues people throw up and at the same time look at better visualisation for people than we had before.

++1

In reply to a post by Malwaremike:

I've just found the new TBB speed tester and I'm greatly impressed. Congratulations to those who developed it and thanks for making it available to us.

Yes, a nice improvement. Well done to all involved

Seems to work on fast connections now too, but still less than speedtest.net says

http://www.thinkbroadband.com/speedtest/button/13611...

Yes its very nice Mr S and its good to see a raw speed figure minus HTTP overhead along with an HTTP figure.

Can I ask wheher an HTML 5 version is in the works at some point (I ask purely out of interest, not because I have anything against Flash)

All I can say is that the release of the flash client is the start of lot of changes

As the resident tame programmer and editor got a lot more things up my sleeve

It can go faster http://www.thinkbroadband.com/speedtest/results.html...

We have seen some interesting things when monitoring network interfaces and testing on high speed connections.

The new speed tester is excellent - just one very minor criticism though: I like to see speeds in Kbps as well as Mbps

Though now it just keeps crashing every time I try it, like with the old version where I kept getting an unexpected error.

Not changed any of the code, so question is where abouts is it stopping?

During

latency checks
Pre test download
Pre test upload
Download
HTTP X2 test
Upload
When about to display results
After results displayed

It was between the 2X upload and download tests, but seems to work ok now.

Good work
As you are tame I'll ask...

It would be great if you could offer an option for people to embed the TBB speed test into your website. Apart from being a useful service to people like me who like to tinker with my website it would also help spread the word about TBB to those that haven't found you!

I have a page that is 1st on Google out of 50+ million results for the term:
how fast is my internet running and similar searches.

But recently the 3rd party speed test service I'm using has proved very erratic
I've been talking with the suppliers and they can see the problem - but things look as if they might take some time to fix.

I would be very happy to use an embedded TBB speedtest - please let me know what you think

There is no 2x upload test

Button embedding will probably come in the next week or two, for now the twitter button gives a very nice short URL

i.e. http://tbb.st/136104491928297589760

For signature or sharing use.

On more commercial arrangements - that would be a talk to Seb, but most likely a smaller one more suited to embedding may appear.

Keep up the excellent work.

sorry, round the wrong way, 2x download, and then it crashes. before the upload. It's doing it at the moment.

It can go faster .......

Yes but it's good to see it can handle slow 20CN speeds too

http://www.thinkbroadband.com/speedtest/button/13611...

Nice results page BUT
wont run before 7pm on my Vivaciti Surfwise Variety connection, just stalls at stage 2.
I assume this is something to do with Vivaciti's throttling of P2P traffic during office hours ???
Can anyone explain what is happening /

Post deleted by djfunkdup

We have run the tester on some very slow and intermittent connections and had no problems

So if your ISP is blocking it, i.e. it works at other times of the day then its down to them to fix.

Sorry want suggesting there was a fault with the TBB speedtest, but what mechanism may Vivaciti be using to block this speedtest and not others ?

I will raise a ticket with them but would appreciate what I should be reporting beyond "this site doesnt work during the day".
Are there any other sites which might have similar problems I can try?
Can anyone else confirm the site doesnt work with Vivaciti Surfwise during Office hours ?

Edited by rarrar (Fri 22-Feb-13 23:06:36)

If you get past the TCP latency check then I would normally expect the test to run.

The download is nothing that special just a transfer of data over TCP port 8095, but probably an unrecognised protocol, but that should represent no real problem.

Can you confirm which times the test does actually run?

http://www.thinkbroadband.com/speedtest/results.html...

Test during office hours on Vivaciti Surfwise, graph curves seem better during these times.

A bit of history. When I first signed up to the Surfwise package well over a year ago now, during the day ITV Player and 4OD were impossible to use but BBC iPlayer was fine. Vivaciti NOC "tweaked my profile" and it was fixed and has been ever since. ITV Player and 4OD were downloading in the order of kilobits, so impossible to use but iPlayer and diagnostics were in the order of megabits. Speedtest protocol being treated similarly?

Tested on BT 21CN:

http://www.thinkbroadband.com/speedtest/button/13616...

I must say that is one of the best looking speedtesters I've seen in my time MrSaffron!
Loving the throughput line on the graph showing any fluctuation in throughput.

My only questions are; When is embedded support coming for this awesome new speedtester here on the forums?

And how long does it take for data from the new speedtest platform to get onto the broadband speed map?

Edited by chris6273 (Sat 23-Feb-13 08:30:26)

They may be throttling us back, but even if down to 20 Kilo bits per second I would expect it to run, just take a very long time

Thank you for the compliment

In terms of forum embedding - in the future
Before that
1. Pushing results to the maps
2. Letting people link it to the connections in the thinkbroadband.com and displaying on the graph

Are the results on the maps all from the old speedtest then?

Currently yes that will change though

Yes, same with my system. On Speedtest.net it regularly reads 900Mb+ in both directions (best ever was 944 in both directions) but the TBB test clocks in at around 200Mb.
All I'm measuring is the speed of the test server, not my fiber connection.
As far as we can make out there is only one testing server (Okola beta in Amsterdam) that can keep up with 1Gb symmetrical fiber connections.
Does anyone know another?

Have you actually looked at the throughput that ookla is generating and compared that to the numbers it displays?

A big number at the end of the test is often not the full story.

We are seeing a handful of people who do have access to fast connections giving reasonable results e.g.

775 Mbps ipv6 TalkTalk
712 Mbps Hyperoptic
497 Mbps Warwick uni
440 Mbps Oxford uni
416 Mbps Essex uni
388 Mbps Ask4

This is not tests by us, but real people. It would be interesting if you could post a link to your results button, as the graph may actually match what speedtest.net would display, I know how their algorithms work so can usually suggest what they would have displayed.

Well I'm running the TTB Meter at the same time as the Ookla / Speedtest.net test and the figures seem to agree pretty well.
Speedtest says 912.66 Mb/s / 936.46 Mb/s and the bar graphs are peaking at just under 1Gb.
TTB is a great tool with so many bells and whistles, is there any way of grabbing the results in order to post them here that doesn't involve an image?
The "Details of Sites Visited" page is quoting speeds of 800 to 1000 KB/Sec after I've run these tests, but that doesn't seem right, this is only about 7-8 Mbps by my reconing, even you own test tells me 200+Mbps
http://www.thinkbroadband.com/speedtest/button/13633...
It's slightly confusing to have both bits/sec and bytes/sec, the conversion site I'm using seems to equate 1Bps=7.8bps approximately.

Edited by deleted (Fri 15-Mar-13 16:57:57)

1 Byte comprises 8 bits

Bytes is used when talking about files,
bits when referring to throughput/connection speeds etc

There is the ability to run a 30 second TCP data grab, but it can be involved to reverse engineer the speed from that. I would rather you had posted the button with the graphs on it, as those are much more useful.

The in and out of the network we are running on is well capable of Gigabit speed, the big unknown is what networks are involved between you and us and that might be the key to the puzzle. As from the numbers I can see you uploading fine, but there are signs of congestion on the download side.

In reply to a post by pjaj:

It's slightly confusing to have both bits/sec and bytes/sec, the conversion site I'm using seems to equate 1Bps=7.8bps approximately.

There are two major confusions, and one sort of fact.

The fact is that for all the discussions here there are 8 bits to a byte. (In systems other than any used here there can be various numbers of bits to a byte).

The two major confusions are to do with the difference between data storage capacity and data transmission speeds

Data storage is measured in bytes, with 1 kilobyte = 1024 bytes, and 1 megabyte = 1024 kilobytes. Independent of the number of bits in a byte.

Data transmission speeds are measured in bits per second, with 1 kilobit ps = 1000 bits ps and 1 megabit ps = 1000 kilobits ps.

(There is further minor confusion about the usage of kilo and kibi as prefixes).

Where things really come to grief is that file download software normally reports in the number of "storage" bytes of the transmitted data. Which makes sense as the downloader is normally interested in how many bytes have been transmitted and remain from his 574 megabyte file. This also caters for the other systems mentioned earlier, where there are other than 8 bits to a byte.

Comparing file download speeds with data transmission speeds is therefore to be done with care. The conversion site you refer to with 1MBps = ~7.8Mbps may be mixing things up.

I've replaced the link with one to the graphs.

Not from what I see you haven't . Which isn't surprising seeing as your post with the link hasn't been edited.

Edited by RobertoS (Fri 15-Mar-13 16:54:32)

Wierd! I've edited it 2 or 3 times now and it keeps flipping back to the wrong button after I say "change this post"
If it still is wrong, the link is
http://www.thinkbroadband.com/speedtest/button/13633...

Can you do a tracert speedtest4.thinkbroadband.com

I don't know the Gigaclear network arrangement off the top of my head. It is possible they have ensured the route to amsterdam is clear, but not worried about ours before as we never claimed we could test Gigabit speeds until a few weeks ago.

The flat HTTP line suggests there is a limit of some sort somewhere.

Tracing route to speedtest4.thinkbroadband.com [80.249.107.221]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms router.domain_not_set.invalid [192.168.1.254]
2 4 ms 3 ms 3 ms 46.227.147.254
3 7 ms 4 ms 4 ms te2-8.ccr02.lhr01.atlas.cogentco.com [149.6.8.201]
4 5 ms 5 ms 5 ms te0-0-0-3.mpd21.lon13.atlas.cogentco.com [130.117.50.217]
5 6 ms 5 ms 5 ms te0-4-0-1.ccr21.lon01.atlas.cogentco.com [130.117.0.57]
6 6 ms 6 ms 5 ms te0-6-0-2.ccr21.lon02.atlas.cogentco.com [130.117.1.218]
7 5 ms 5 ms 5 ms te2-3.mag01.lon02.atlas.cogentco.com [130.117.50.118]
8 5 ms 5 ms 5 ms cgt-gw1.thdo.ncuk.net [149.6.148.114]
9 5 ms 23 ms 6 ms te2-1-9.star10g.bdr-rt1.thn.ncuk.net [80.249.97.18]
10 5 ms 5 ms 5 ms ip107-221.rbsov.ncuk.net [80.249.107.221]

Trace complete.

In reply to a post by pjaj:

the conversion site I'm using seems to equate 1Bps=7.8bps approximately.

That's impossible! An (8-bit) Byte sitting in your PC would need more than 8 bits to be transmitted, cuz of the TCP/IP overheads that were carried with it. So 1 Bps arriving at the PC > 8 bps on the wire. Maybe they subtracted when they should have added and mean 1 Bps = 8.2 bps.

UNLESS they they are trying to average over all the diff Byte lengths. But the is extremely unlikely as the 8-bit Byte massively predominates over the Internet.

Ta have sent someone off to take a look

As far as we know, Gigaclear are not deliberately throttling the traffic or tailoring it in any way.
They claim each subscriber has 10Mb minimum guaranteed at all times or as much of the 1Gb back haul speed as is free, which could be all of it if no-one else is on line.
The back haul feed goes about 3-4km cross country to join up with the C&W fiber backbone and thence straight down to their Linx exchange in London.
It bypasses all BT networks.

In reply to a post by XRaySpeX:

In reply to a post by pjaj:

the conversion site I'm using seems to equate 1Bps=7.8bps approximately.

That's impossible! An (8-bit) Byte sitting in your PC would need more than 8 bits to be transmitted, cuz of the TCP/IP overheads that were carried with it. So 1 Bps arriving at the PC > 8 bps on the wire. Maybe they subtracted when they should have added and mean 1 Bps = 8.2 bps.

UNLESS they they are trying to average over all the diff Byte lengths. But the is extremely unlikely as the 8-bit Byte massively predominates over the Internet.

I agree, but don't shoot the messenger! I'm only reporting what a site offering conversions says.
In my day 1 byte was always 8 bits, but when wrapped for transmission it went up to 10 or 11 bits.
Yes, I started life using a teleprinter or punched cards to communicate with computers!
VDUs - luxury!

That wasn't a byte. It was a character, and you were dealing in bauds .

In reply to a post by RobertoS:

That wasn't a byte. It was a character, and you were dealing in bauds .

I was simplifying.