Speed test sites and why you shouldnt take them for granted
21-02-2012, 19:45
|
#61
|
Inactive
Join Date: Aug 2005
Location: Scotland
Age: 42
Services: Virgin Media - XL Plus package with XXL broadband
SKY HD Multiroom
Freeview HD
Freesat HD
Posts: 2,816
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by roughbeast
BINGO! You got it. This is why non-experts like me come to this forum and try to get explained what seems inexplicable, hopefully without being patronised.
So far the explanations I have received haven't answered the key question:
How can TBB base a national survey of comparative ISP speeds on a monitoring tool that limits its devices to a single thread speed test that seems incapable of dealing with congested times of the day?
Indeed with a maximum test speed of 42Mb at one point in a 24 hour period it hasn't managed to deal with low congestion.
Do they take my average TBB result of 27Mb and multiply it up, as has been suggested, by a factor of 4 or do they just conclude that my 100Mb can only manage an average of 27Mb? ( Between 18Mb and 42Mb over 5 tests ) After all, the point of the survey, as I understand, it is to see if ISPs give customers the speed they advertise.
BTW. Sorry about the 16MB. That was a typo. I meant 12MB. I must have been dreaming about the 'Proof of Concept' trial.
|
There is no speed test that will be able to truly give a single thread test no matter what tiem of the day, no one has unlimited bandwidht or should i say backhual ie line speed intot he server which i suspect is gbit or 10gbit for thinkbroadband with peopel getting faster and faster conention they would ned to match that upgrade in speed but dnt or cant oen of the two
end of the day tbb can measure yoru speed of 100mb on single thread but it has to be when they are suffering congestion no one can get uncongested networks unless they pay for it and that costs millions because oyu need seprerate server runnign on ther eown decicated lines
does that explain it i hope i havent been partonaising
|
|
|
21-02-2012, 20:01
|
#62
|
cf.mega poster
Join Date: Jul 2008
Location: Coventry
Services: Vodafone/City Fibre Gigafast 900
Posts: 1,781
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by Andrewcrawford23
There is no speed test that will be able to truly give a single thread test no matter what tiem of the day, no one has unlimited bandwidht or should i say backhual ie line speed intot he server which i suspect is gbit or 10gbit for thinkbroadband with peopel getting faster and faster conention they would ned to match that upgrade in speed but dnt or cant oen of the two
end of the day tbb can measure yoru speed of 100mb on single thread but it has to be when they are suffering congestion no one can get uncongested networks unless they pay for it and that costs millions because oyu need seprerate server runnign on ther eown decicated lines
does that explain it i hope i havent been partonaising
|
No, you you are never patronising.
You have answered my question though. Inability to counteract congestion for testing purposes seems to be a fatal flaw. TBB cannot use the monitoring tool installed on my PC, and thousands of others, as the basis for a national survey of comparative ISP speeds. Does anyone actually disagree with this conclusion? If so, on what basis?
__________________
Join Date: Jul 2008
Location: Coventry
Services: FACTCO/CityFibre 1GB FTTP; Asus GT-AX11000 +3 iMesh nodes; Humax 2Tb TV boxes x2; Synology DS920+ used as Plex server
|
|
|
21-02-2012, 20:13
|
#63
|
Inactive
Join Date: Aug 2005
Location: Scotland
Age: 42
Services: Virgin Media - XL Plus package with XXL broadband
SKY HD Multiroom
Freeview HD
Freesat HD
Posts: 2,816
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by roughbeast
No, you you are never patronising.
You have answered my question though. Inability to counteract congestion for testing purposes seems to be a fatal flaw. TBB cannot use the monitoring tool installed on my PC, and thousands of others, as the basis for a national survey of comparative ISP speeds. Does anyone actually disagree with this conclusion? If so, on what basis?
|
i agree the program in it current format cant be used for national isp statis
however if they changed the program to work liek this..
firs tthey find out your advertised conneciton speed, when th test runs it checks the serve for ulsations ie they maybe know they have 1gbit connection, so it check ok we have 5 100mb users on we have neough bandwdiht to do another etst and runs the test.
however if it checks and find it currently has 9 100mb user a 50 mb user a 20mb user and it wants to runanotehr 100mb test it wait or some of the test to finish so there neough bandwidht to test it sand then it runs but it also makes sure no one else can connect until ther eis enough bandwidht so use first come first sevrer piraioty system
|
|
|
22-02-2012, 04:13
|
#64
|
cf.mega poster
Join Date: Nov 2010
Location: Stafford
Posts: 4,226
|
Re: Speed test sites and why you shouldnt take them for granted
there has always been a routing error between TBB and Virgin Media users, nothing new and nothing to do with the speed test itself, mind you virgins routing to most servers is pretty crap
|
|
|
22-02-2012, 05:05
|
#65
|
cf.mega poster
Join Date: Sep 2003
Services: Gig1, Hub 5
Posts: 12,043
|
Re: Speed test sites and why you shouldnt take them for granted
100mbit is easily doable in a single thread. I have done transfers at gigabit speeds over single threads between servers.
Its just harder to do when rtt increases. However I would expect it possible from a uk server to a uk connection.
so whilst the threading doesnt help it doesnt mean thats a test breaker.
|
|
|
22-02-2012, 14:54
|
#66
|
cf.mega poster
Join Date: Nov 2010
Location: Stafford
Posts: 4,226
|
Re: Speed test sites and why you shouldnt take them for granted
most speed test problems on VM connections are routing, there are UK servers capable of 500mbit+ but can only manage about 30/40mbit from a VM line, no matter what time of day, this has been proved by running a test to the maidenhead server on speedtest from a VM connection, then from a gigabit web server
|
|
|
22-02-2012, 15:17
|
#67
|
cf.mega poster
Join Date: Aug 2004
Posts: 11,207
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by roughbeast
BINGO! You got it. This is why non-experts like me come to this forum and try to get explained what seems inexplicable, hopefully without being patronised.
So far the explanations I have received haven't answered the key question:
How can TBB base a national survey of comparative ISP speeds on a monitoring tool that limits its devices to a single thread speed test that seems incapable of dealing with congested times of the day?
Indeed with a maximum test speed of 42Mb at one point in a 24 hour period it hasn't managed to deal with low congestion.
Do they take my average TBB result of 27Mb and multiply it up, as has been suggested, by a factor of 4 or do they just conclude that my 100Mb can only manage an average of 27Mb? ( Between 18Mb and 42Mb over 5 tests ) After all, the point of the survey, as I understand, it is to see if ISPs give customers the speed they advertise.
BTW. Sorry about the 16MB. That was a typo. I meant 12MB. I must have been dreaming about the 'Proof of Concept' trial.
|
Congestion is part of your connection. "Dealing with it" only makes your connection seem faster than it actually is.
Like I've said before, if you want to know the maximum speed your connection is possibly capable of if you rule out all other factors, you shouldn't be doing a speed test to begin with. Go look at your modem status page or your bill where it says which internet service you subscribe to.
If you want to know what speed your connection actually attains when doing something meaningful, such as fetching a single file, then the TBB speedtest is far more accurate. All a speedtest does is download a file for you and do the counting for you, nothing more. And that's all it's supposed to do.
For your question, which seems to be completely irrational:
Quote:
How can TBB base a national survey of comparative ISP speeds on a monitoring tool that limits its devices to a single thread speed test that seems incapable of dealing with congested times of the day?
|
The only thing a speedtest can be used for is to measure your speed WITH congestion. If your connection can't run at full speed because of congestion, that's EXACTLY what they're designed to show.
Are you suggesting a comparative survey of ISP speeds should only look the maximum theoretical attainable speeds of your connection, under perfect conditions, and without any external influences? If so, it's easier just to ask what tariff you subscribe to. No need to test anything.
So, again, what do you find out from a speedtest? Like I've said several times in this thread:
1) If you want to know the maximum possible speed of your connection, look at your bill or your modem status page.
2) If you want to know the speed you can download a file from a certain place at a certain point in time, do a speedtest.
If a speedtest slows down because of congestion that is exactly its job. If a speed test tries to deal with congestion and hide it, it's trying to do the wrong job and doing it badly.
---------- Post added at 15:10 ---------- Previous post was at 15:05 ----------
Quote:
Originally Posted by Andrewcrawford23
There is no speed test that will be able to truly give a single thread test no matter what tiem of the day, no one has unlimited bandwidht or should i say backhual ie line speed intot he server which i suspect is gbit or 10gbit for thinkbroadband with peopel getting faster and faster conention they would ned to match that upgrade in speed but dnt or cant oen of the two
|
Incorrect. Most single thread tests get within 90% of my max speed most times of day. Only on VM is it unlikely, and even then only on higher tariffs. Take a look at the OFCOM reports where they compare single vs. multithreaded speeds. VM sees the biggest loss in speed going from multithreaded to singlethreaded but the drop is on average usually less than 20%. Most other ISPs lose less than 10%.
---------- Post added at 15:13 ---------- Previous post was at 15:10 ----------
Quote:
Originally Posted by roughbeast
You have answered my question though. Inability to counteract congestion for testing purposes seems to be a fatal flaw. TBB cannot use the monitoring tool installed on my PC, and thousands of others, as the basis for a national survey of comparative ISP speeds. Does anyone actually disagree with this conclusion? If so, on what basis?
|
I completely disagree.
Not dealing with congestion is the whole point of speedtests. It's not a flaw, it's doing it's intended job. Even if that's not the job you think or want it to do. The only fatal flaw is your idea of what a speed test is for.
Their national survey of comparative ISP speeds aims to compare actual attainable speeds, not theoretical maximums.
This is exactly what the TBB speedtest does and it does what it's designed for. Speedtest.net already excels at doing a pretty half-arsed job at comparing theoretical maximums, we don't need another wannabe site trying to follow suit.
There is no point comparing people's speeds if you hide everything that affects your speeds (i.e. primarily congestion). Everyone would get exactly the same as the speed their modem says they get.
---------- Post added at 15:14 ---------- Previous post was at 15:13 ----------
Quote:
Originally Posted by craigj2k12
there has always been a routing error between TBB and Virgin Media users, nothing new and nothing to do with the speed test itself, mind you virgins routing to most servers is pretty crap
|
Routing between TBB and Virgin Media is much better than average.
---------- Post added at 15:17 ---------- Previous post was at 15:14 ----------
Quote:
Originally Posted by craigj2k12
most speed test problems on VM connections are routing, there are UK servers capable of 500mbit+ but can only manage about 30/40mbit from a VM line, no matter what time of day, this has been proved by running a test to the maidenhead server on speedtest from a VM connection, then from a gigabit web server
|
If a speed test shows you have a low attainable speed between your location and a another location, this shows your ISP has a poor route between those locations.
The only problem there is your ISP, not the speed test. The speed test is doing its job perfectly.
|
|
|
22-02-2012, 16:35
|
#68
|
Ran Away
Join Date: Nov 2008
Location: Lincoln
Services: phone + 1gbit BB + SkyQ
Posts: 11,021
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by craigj2k12
most speed test problems on VM connections are routing, there are UK servers capable of 500mbit+ but can only manage about 30/40mbit from a VM line, no matter what time of day, this has been proved by running a test to the maidenhead server on speedtest from a VM connection, then from a gigabit web server
|
what is so screwed up with the routing that it is impacting speedtest results?
|
|
|
22-02-2012, 16:48
|
#69
|
Inactive
Join Date: Aug 2005
Location: Scotland
Age: 42
Services: Virgin Media - XL Plus package with XXL broadband
SKY HD Multiroom
Freeview HD
Freesat HD
Posts: 2,816
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by qasdfdsaq
Incorrect. Most single thread tests get within 90% of my max speed most times of day. Only on VM is it unlikely, and even then only on higher tariffs. Take a look at the OFCOM reports where they compare single vs. multithreaded speeds. VM sees the biggest loss in speed going from multithreaded to singlethreaded but the drop is on average usually less than 20%. Most other ISPs lose less than 10%.
|
not in the form i mean it, i meant that statement in teh form that since no speed test i know works on teh basis of limiting users takign the test deendent ont eh cjonnection spee dof the server testing and the amount of user conection conencting then no single thread one will ever give a true speed, i dnt doubt it can i have gotten the proper speed forma single thread test before but as you has said it down to congestion and that is wha ti was saying ina weird way, but nto jsut ocngestion on virign line and backhual but also the speed test site line
|
|
|
22-02-2012, 17:54
|
#70
|
cf.mega poster
Join Date: Aug 2004
Posts: 11,207
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by Andrewcrawford23
not in the form i mean it, i meant that statement in teh form that since no speed test i know works on teh basis of limiting users takign the test deendent ont eh cjonnection spee dof the server testing and the amount of user conection conencting
|
As this was the only part I could understand, even if only barely:
Also incorrect. There are many speed test sites that limit users taking the test based on connection speed and server capacity. Perhaps most don't but there are definitely many that do.
|
|
|
22-02-2012, 18:36
|
#71
|
Inactive
Join Date: Aug 2005
Location: Scotland
Age: 42
Services: Virgin Media - XL Plus package with XXL broadband
SKY HD Multiroom
Freeview HD
Freesat HD
Posts: 2,816
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by qasdfdsaq
As this was the only part I could understand, even if only barely:
Also incorrect. There are many speed test sites that limit users taking the test based on connection speed and server capacity. Perhaps most don't but there are definitely many that do.
|
i never said there wasnt :p jsut none that i am aware of, i prefer speed test sites to work that way means you knwo they have the bandwidht to test you nd give a fairer test of your connection but still not the be all end all
|
|
|
22-02-2012, 20:13
|
#72
|
cf.mega poster
Join Date: Nov 2010
Location: Stafford
Posts: 4,226
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by General Maximus
what is so screwed up with the routing that it is impacting speedtest results?
|
dont know, i imagine its the fact that VM try to put so many users bandwidth through such a small backbone
---------- Post added at 20:13 ---------- Previous post was at 20:12 ----------
Quote:
Originally Posted by qasdfdsaq
The only problem there is your ISP, not the speed test. The speed test is doing its job perfectly.
|
yeah thats what im saying, the speed tests are reliable and its at VMs end where the problem exists
|
|
|
22-02-2012, 20:18
|
#73
|
cf.mega poster
Join Date: Aug 2004
Posts: 11,207
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by craigj2k12
yeah thats what im saying, the speed tests are reliable and its at VMs end where the problem exists
|
OK, I must have misunderstood then.
You did say "most speed test problems" rather than, say "most low speed problems" or "most connection problems".
|
|
|
23-02-2012, 12:43
|
#74
|
cf.mega poster
Join Date: Nov 2010
Location: Stafford
Posts: 4,226
|
Re: Speed test sites and why you shouldnt take them for granted
Quote:
Originally Posted by qasdfdsaq
OK, I must have misunderstood then.
You did say "most speed test problems" rather than, say "most low speed problems" or "most connection problems".
|
yeah i meant speed testing from VM connections, my bad
|
|
|
23-02-2012, 20:46
|
#75
|
Sulking in the Corner
Join Date: Jul 2009
Location: RG41
Services: 1 Gbps; Hub 4 MM; ASUS RT-AX88U; Ultimate VOLT. BT Infinity2; Devolo 1200AV
Posts: 11,955
|
Re: Speed test sites and why you shouldnt take them for granted
I don't know how this discussion has backed itself so far into a corner. Speedtests on the VM network are influenced by:
- Congestion (if any) at the optical node
- Congestion (if any) at the line card
- Congestion (if any) at the CMTS itself (unlikely)
- Congestion (if any) in the core (rare but not unknown)
- Congestion (if any) in the VM backbone (rare and occurs if there is a routing problem elsewhere)
- Congestion (if any) at the peering point (deffo possible)
- Congestion (if any) at the speed test server (e.g. Maidenhead - so well known)
Simples.
__________________
Seph.
My advice is at your risk.
|
|
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT +1. The time now is 03:41.
|