Which one to believe?

[saurabhdua]

Hello Folks!

 

Please refer to the attached comparison & help clear the dilemma associated to the veracity of these different Speed-tests!?

Which one to be believed & showcased to my ISP as  Proof?

 

My ISP charge me Rs.1034/- (approx. $21) per month for a 50Mbps Speed package.

 

& Lastly, whether Speed-Tests  are the only way to prove a point to the ISP?...or, even more extensive tests are available to surface the Network-Health, Latency etc. etc. ?

 

Please suggest further in this regard.

 

Thanks.

[T3X4S]

OK - I havent seen THAT much of a difference between fast.com & speedtest - thats pretty big.

When testing, are you testing via wireless ?  Need to do it hardwired if you are testing the ISP's speed claims - have to rule out as much outside influence as possible.

Also, if these tests are hardwired - then it doesnt matter - because neither is anywhere close to the 50Mbps they are claiming - Im sure like in US they say "up to 50mbps" - but that isnt close.

 

Sure there are more technical tests - but you are going to be dealing with a low end customer service monkey - they wont understand if you threw a bunch of wireshark logs at them and showed them extreme packet loss... they dont care, dont understand, and dont get paid enough to care.


If your IP has their own speed test site - or a module on the router - that would be helpful.

BUT - in lack of a perfect testing tools - you can hit them with 5 imperfect testing tools - 

If you have 5 web-based speed tests - and all show a speed way below what is claimed - that is good enough.

Remember, the goal is
(1) for them to send a tech out to your house and run some onsite tests to see if there is a problem & fix it
(2) Give you a discount due to them not being close to their claims and mention things "false advertisement", "coercion", "fraud", "conspiracy to commit fraud", hell blame them for JFK's assasination - it doesnt matter

(3) To help you in someway to be a satisfied customer

Your goal isnt to get the person mad, - remember -  they dont care - they aren't paid enough to care.

" ...help clear the dilemma associated to the veracity of these different Speed-tests .... " - now I know you aren't using Google translate -             sorry - the verbiage just sounds funny
 

[+BudMan MVC]

where is the netflix test to you?, and that 2nd one looks like huge amount of latency as well.  So that would explain lower values.

 

But if that speedtest one is half of what your suppose to get.. Then yeah something is not correct..  Speedtest will use up to like 8 streams I believe to try and max out your pipe.  If  you can not get what your paying for with that test then there is a problem.

 

That should be more than enough to complain to your isp with.  You know if you were seeing 48 and you paid for 50, prob not going to amount much.  But 27 and your paying for 50..  And that is the best you ever see - even odd ball times outside of prime usage times, etc.?

[saurabhdua]

37 minutes ago, T3X4S said:

Also, if these tests are hardwired - then it doesnt matter - because neither is anywhere close to the 50Mbps they are claiming - Im sure like in US they say "up to 50mbps" - but that isnt close.

Hello T3X4S!

 

Iam always  Hard-wired to an Internet connection & NEVER use a Wi-Fi ! The extreme dips are more evident during Peak-Hours. 

Please refer to the http://bit.ly/2dbwAMu & help ascertain could that be the cause behind this 'Compromised' performance?

[saurabhdua]

33 minutes ago, BudMan said:

where is the netflix test to you?

Hi BudMan!

 

I couldn't get this point? Netflix's owned Fast.com does the required automatically!

barring Peak-hours, 'Speedtest.net' is almost consistent to depict -- "All is Well" kinda status!..& mostly showcase Speeds near to Perfect score.

 

How can the Peak-Hours lacunae be Highlighted to my ISP? Please suggest? Should I start compiling Data on a weekly basis & portray a consolidated record to facilitate an improvement in this context?

[Hum]

My cable modem speeds vary a lot.

 

Depends on time of day, and whether things need a reboot.

 

Speeds can be a low 3+ Megs to 74+ Megs.

[saurabhdua]

42 minutes ago, BudMan said:

But 27 and your paying for 50..  And that is the best you ever see - even odd ball times outside of prime usage times, etc.?

"Speedtest.net" generally clocks the 'Perfect' score most of the times!...but deviates occasionally during Peak-Hours (Night-Hours). As far as Test from Netflix is concerned, 40Mbps is the BEST I have observed thus far! & that too during only Off-Peak hours!

 

Would you mind scouring http://bit.ly/2dbwAMu & thus suggest if this could possibly the cause behind such a 'Compromised' performance?

 

Inputs will be greatly appreciated. Thanks.

[Hum]

If you get a lower speed than 50 Mbps from a particular web site, that is simply normal.

 

It all depends on traffic between that one web site, and what speed they can send.

 

Just because your ISP advertises a possible 50 Mbps does not mean that every Server is sending at that speed.

 

Changing your TCP settings and DNS servers can also speed up your Net reception.

[saurabhdua]

Its 1:30 AM of 27th morning & the Latest attempt have fetched me these results.

 

However, I would still  request the Members to help me explain the analogy ###### implications associated to the hive of these Critical errors & Warnings showing up under - http://bit.ly/2dbwAMu

 

What sort of anomaly does this skewed SNMP Log suggest?

 

 

[saurabhdua]

8 minutes ago, Hum said:

Changing your TCP settings and DNS servers can also speed up your Net reception.

Hello Hum!

 

I snub my ISP's DNS always & use Google DNS  (or sometimes OpenDNS) instead!

Please reveal more towards TCP settings? How can these be maneuvered for a purpose? 

[Hum]

I use TCP  Optimizer and DNS Jumper 2.

 

You'll have to do a Search for those programs and Download them.

 

Remember, the advertised ISP speeds are 'up to' -- which is a catch phrase.

 

You will not always get 50 Mbps, no matter what.

[+BudMan MVC]

those errors have nothing to do with your issue.  So speedtest always shows you 50 or close to 50, and your wondering why you don't see 50 on fast.com?  Is that the question?

 

Is so is because fast.com is how far away from you?  Its only 1 stream..  You could sniff the traffic to find the IP your testing to when you run fast.com - but I highly doubt its going to show you the same speed speedtest.net does.

 

Especially if your in some remote part of the world.  I juts did a test for fast.com and got 36Mbps, while speedtest.net shows me getting 80mbps - I pay for 75/10..

[sc302 Veteran]

different servers will provide different results.  Even on speed test, it will try to find the fastest server closest to you to connect to and perform the test....many times, if you manually select a server somewhere else, it may return faster or slower speed tests. 

 

You are connecting to the netflix servers to get the fast speed, and because they are different than the speedtest servers, the speeds may not coincide. 

 

 

FWIW, I am getting between 76-80 on both fast and speedtest.net...but that is of little difference to the matter.

[xendrome]

3 hours ago, saurabhdua said:

Hello Hum!

 

I snub my ISP's DNS always & use Google DNS  (or sometimes OpenDNS) instead!

Please reveal more towards TCP settings? How can these be maneuvered for a purpose? 

DNS and TCP settings aren't going to speed up these types of usages/tests. Or actually anything in the real world, DNS maybe a little... but not the other.

[Pork Chopper]

If this was mentioned then I'm sorry for repeating it.

 

A site like speedtest.net measures speed based on the speed of the data received.  It does not use any of the TCP/IP overhead bytes in it's final number.  It basically downloads a file of a known size and when it's done downloading it measures the time it took.  So this speed test is more indicative of real life.  It it took into account the TCP/IP overhead that would add about 4-5% to the Mbps.  Other speed tests might include the TCP/IP overhead.  So comparing two speedtests might inherently give you a different final number.

[saurabhdua]

7 hours ago, sc302 said:

different servers will provide different results.  Even on speed test, it will try to find the fastest server closest to you to connect to and perform the test....many times, if you manually select a server somewhere else, it may return faster or slower speed tests. 

Hello sc302!

 

That's how I have come to realize right now that why the "File Transfer Rates" doesn't always commensurate with Connection Speeds!

For instance, a Connection Speed of say 48Mbps should ideally fetch me 'Transfer rates' of 6MB/sec {48/8}..but in actual it hovers around 3.5 to 4MB/sec , & that too when using some kind of Download-Accelerator!

 

So basically its all illusion! Well...Are there some ISPs in your part of the World which do Guarantee a near Perfect score at mostly all times of the day?

 

& I wonder...if there ever would be an analogy of say - "Smart ISP" in the time to come....capable of maneuvering the Download & Upload Speeds on Demand!?

[saurabhdua]

May I please request the Dear Members once again to please scour the revelations posted against http://bit.ly/2dbwAMu & help derive possible implications on the overall performance?

 

Are these Critical Warnings & Error messages signify some major lacunae at my ISP's end? 

 

Inputs are keenly awaited....

[sc302 Veteran]

2 hours ago, saurabhdua said:

Hello sc302!

 

That's how I have come to realize right now that why the "File Transfer Rates" doesn't always commensurate with Connection Speeds!

For instance, a Connection Speed of say 48Mbps should ideally fetch me 'Transfer rates' of 6MB/sec {48/8}..but in actual it hovers around 3.5 to 4MB/sec , & that too when using some kind of Download-Accelerator!

 

So basically its all illusion! Well...Are there some ISPs in your part of the World which do Guarantee a near Perfect score at mostly all times of the day?

 

& I wonder...if there ever would be an analogy of say - "Smart ISP" in the time to come....capable of maneuvering the Download & Upload Speeds on Demand!?

See if the isp has their own speed test.  If they do, try that. If you are at speed then all you can do is switch providers in hopes that they have a better connection to the world. 

[saurabhdua]

3 minutes ago, sc302 said:

See if the isp has their own speed test.  If they do, try that.

Well no! Their website -- http://bit.ly/2dyVLdV do not host any such option!

Did you also bother a keen look at http://bit.ly/2dbwAMu to enumerate over possible implications one could experience under such a scenario?

 

Your most valuable Inputs are being awaited on the Subject. Thanks. :-)

[sc302 Veteran]

Yes. Irrelevant noise. 

[PsYcHoKiLLa]

20 minutes ago, sc302 said:

See if the isp has their own speed test.  If they do, try that. If you are at speed then all you can do is switch providers in hopes that they have a better connection to the world. 

He does have an overlay in the first screenshot that Speedtest.net is downloading from his ISP's own servers so that's pretty much the same, no?

[sc302 Veteran]

2 minutes ago, PsYcHoKiLLa said:

He does have an overlay in the first screenshot that Speedtest.net is downloading from his ISP's own servers so that's pretty much the same, no?

Even though comcast does have speed test servers, I do get better results with the direct comcast speed test servers as an example of this. 

 

http://speedtest.xfinity.com

[+BudMan MVC]

Thinking that size of the pipe the ISP gives you 50mbps is going to = what your download speed all the time is not correct.

 

For starters you have to take into account the latency of where your downloading.  Even if your pipe was 10GBps tcp doesn't work that way!!  How many streams, what is the latency.. What is the receive window, etc..

 

So if your downloading from a server that is 10ms a way vs a server that is 100ms away going to have a HUGE difference on your actual speed.

 

Do the math!!

So your pipe is 50mbps, lets say your 10ms away with default 64k window size.  You could in theory get 52.43 Mbit/sec. so sure you could fill you pipe

No lets say your 20ms away..

1 stream is going to only get you 26.21 Mbit/sec.  You would have to bump you windows size up to 122KB to max out your pipe

No lets say your 40ms away..

Your talking only 13.11 Mbit/sec. You would need 244KByte windows size to max out your 50mbps

 

So does not always matter how FAT the pipe is but how long it is... So how far away fast.com they do not even tell you a ping?  But I doubt they are next door to you..   Speedtest gives you how far away with the ping..   Do a speedtest from a sever far away compared to close.

 

So here is some speed test history.. Notice all chicago, close - 10 ms.. Maxing out my pipe I pay for 75/10

 

But if I pick a server far away - see 68ms to this server in CA, I don't see my 80mbps do I.

 

So lets do the math.. At 68ms away with 1 stream and 64KB window your only going to see 7.71 Mbit/sec.  But I see more than that in the test because they are using more than 1 stream and or larger window size..  If you check out their FAQ they can use up to 4 streams.. You would have to sniff the traffic to see what window rcv size they are using..  But if you do 7.71 x 4 = 30 and was seeing just a bit under that..  So could be just default window size and 4 streams.

 

Your NOT going to see 50mbps all the time from everyone is the point here.  If y our speedtest is showing you what you paid for your prob fine.

 

Here this will explain how latency has effect on your speed you see.

https://en.wikipedia.org/wiki/Bandwidth-delay_product