[SOLVED, by Terrorkarotte] Please help to verify my uplink speed...

[3choE]

Hi guys,

I am confused about my uplink speed and hope you guys can help me out.

For my hosting account I have a 100 mbps uplink. I did some speed tests at speakeasy . net and speedtest . net (I have to put spaces in the URL string, otherwise Akismet would complain), and got the following results:





Yet the tech support staff said those two websites weren't designed for measuring commercial line, and therefore the results weren't accurate. He suggested me to use visualware . com instead, and I got the following:



I have also tried downloading a 500 MB testing file as he suggested, and got the following (from using Firebox, and also Wget on Windows):





(Firefox's download manager is using the unit MB, this is megabyte? How about the unit M that Wget is using?)

The results from speakeasy . net, speedtest . net, and visualware . com don't look like a 100 mbps to me. Then the support staff ran few tests himself locally from my box, and replied me saying that he got ~12 MB/s, so my uplink was good. (I don't know which test that he got the result from. Maybe from Firefox download manager?)

Can someone help me to confirm from the above results, whether I got a 10 mbps, 100 mbps uplink?

---

Also, when I did a test at pingtest.net, I got an average of 60 ms jitter:



Yet my home DSL I got an average of 5 ms jitter:



Does this jitter thing matter? (I checked the definition already, but want to know the real-life affect on a game server)

Again, the tech support staff claims those webtools are not designed for measuring a commercial line, and therefore the results cannot be considered accurate... (But as a home user, those webtools are the one that I have access to... )


Thank you very much.

[loopyman]

Those tests confirm that you're running 100mbps
Yea, I have the same issue with jitter from ping test... My local line shows up as A, and my data center lines result in C's
But my local connection sucks in its reliability.

[Beaverbeliever]

I would say by looking at the results, that your connection has around 10mbps upload.

[SBT]

Using a flash speed test won't accurately show a 100Mbps line, just like it won't show a 1000Mbps line. Try a test download off cachefly.com

[3choE]

loopyman, Beaverbeliever, SBT:

Thanks for the repies.

---

I didn't know where to do the upload test, so I have only done the download, as the down & up of my dedi's should be symmetric.

(Anyone knows any website that offers upload test? By actual uploading a file that is)

File: 64-bit PC (AMD64) server install CD
Size: 654 MB
URL: http://mirror.anl.gov/pub/ubuntu-iso/CDs/9.10/ubuntu-9.10-server-amd64.iso

1)
[wget @ dedi]
~9 min (4:04 pm - 4:13 pm) // ~9 min is the actual time the downloading took, but not from the ETA.


2)
[wget @ home]
~19 min (4:15 pm - 4:34 pm)


3)
[firefox @ dedi]
~9 min (4:21 pm - 4:28 pm)


4)
[firefox @ home]
~21 min (4:38 pm - 4:59 pm)


So the actual down speed is just about 2x faster than my home DSL (which is ~6 mbps down). According to that, I really don't see how my dedi's line can be a 100 mbps...


What do you guys think?

---

Back to my results from Speakeasy (or the other ones):

Dedi:


Home:


while might not be accurate, does match the results I got from the actual downloads (that the dedi's line is about 2.5x faster than my home 6 mbps DSL).

[loopyman]

According to that windows download you have a 10MegaByte Download per second... Which means you have at least 87 mbps... Also if windows is calculating its going to take 22 seconds to download 200MB left of file then it makes sense...

If this is your limestone box, don't worry about it... You have 100mbps up / down...

[3choE]

loopyman,

Thanks again for taking time to help out.

Not trying to be persistence, but I really don't get it...

Maybe I take back a few steps first and come back to this topic another time, so I can see all the bit & byte stuffs more clearly...

---

But before I take my break from this, I am thinking this (let me try the other way):

655 Megabyte / 9 minutes / 60 sec = 1.21 Megabyte per sec = 9.70 Megabit per sec.

---

[Update]
I have found a possible reason...

The 10M/s result, is from downloading the test file that the support staff told me:


I didn't pay attention of the location of that file (LAN?). Yet this afternoon I didn't feel right about it, so I tried to download from a different source. I tried download a Ubuntu .iso file from a couple different mirrors (to be fair about the line condition and the distance), and got the following result:


That probably explains everything. I should check with the support staff again...

[Terrorkarotte]

You know that there are other things than the connection that have a impact of the line and a 100Mbit card never will give you 100MB? 100M BIT = 11.92093 Mega BYTE
Now take in mind that you will not use 100% of the line since all system secure some recources for their networking. If you need more bandwidth go to 1Gbit connection.
Wget shows you Mega BYTE per second. So i guess your connection is fine.

[3choE]

(01-20-2010, 06:01 PM)Terrorkarotte Wrote:  Wget shows you Mega BYTE per second. So i guess your connection is fine.

Terrorkarotte,

Thanks for your input. I got a reply from the support staff, that he is making a similar claim as yours, and therefore refuse to do any further investigation on the issue.

Yet, my concern is, both yours and his suggestion, are based on which of my two results?

This:


or this:

---

(01-20-2010, 06:01 PM)Terrorkarotte Wrote:  100M BIT = 11.92093 Mega BYTE

100 Megabit = 12.5 Megabyte, as it is divided by 8.

---

I am aware of the result is depended on many factors, such as congestion, distance, throughout of the source, and etc, yet the only time I reached ~10 M/s, was when downloading the test file that the support staff suggested. Yet that file was within the same domain, and probably it was a LAN transmission.

As with the other download sources, I have tried about 10 Ubuntu/CentOS mirrors, and the average was:

655 Megabyte / 9 minutes / 60 sec = 1.21 Megabyte per sec = 9.70 Megabit per sec.

Which is ~10% of a 100 mbps connection. I don't see how anyone can call it acceptable.

---

(01-20-2010, 06:01 PM)Terrorkarotte Wrote:  ... that you will not use 100% of the line since all system secure some recources for their networking.

I am not demanding to reach 100% capacity of my 100 mbps connection. I am demanding to reach something better than 10%.

---

(01-20-2010, 06:01 PM)Terrorkarotte Wrote:  If you need more bandwidth go to 1Gbit connection.

So the scheme for realistic performance, would be:

- 1000 mbps for 100 mbps
- 100 mbps for 10 mbps
- 10 mbps for 1 mbps

Correct?

---

I am hoping any member here using 100 mbps (which should be a lot), can kindly do a download from those following Ubuntu/CentOS mirrors and let me know your result:

- http://mirror.anl.gov/pub/ubuntu-iso/CDs/9.10/ubuntu-9.10-server-amd64.iso
- http://ubuntu.cs.utah.edu/releases/9.10/ubuntu-9.10-desktop-amd64.iso
- http://www.gtlib.gatech.edu/pub/ubuntu-releases/9.10/ubuntu-9.10-desktop-amd64.iso
- http://mirror.ash.fastserv.com/pub/linux/centos/5.4/isos/x86_64/CentOS-5.4-x86_64-LiveCD.iso

I believe downloading won't consume your monthly allowance, and a partial download should be fine for getting a rough measurement.

[Terrorkarotte]

It is based on the first result. The second one obviously got a bad server on the other end.
Its not only 100/8 8bit=1byte; 1024byte=1kbyte 1024kbyte=1Mbyte
so 100000000bit:
-1250000byte
-12207.03125kbyte
-11.92093Mbyte
-0.0116415Gbyte

AGAIN M WITH WGET MEANS MEGABYTE!
So you get 91.4% of your theoretical maximum speed if you reach 10.9M/s.

I reach exactly the same with my connections so its seems the effictive maximum for a 100Mbit line

Read the contract. I am sure they only offer you a 100Mbit card that is connected to the internet or so. Not you get 100Mbit internetconnection.

If i were you start two or more downloads from fast mirrors and see how much they reach together.

If it is to less in your eyes change the provider.

The realistic is something above but not much.

[3choE]

Terrorkarotte,

Nice having you back to this thread. I hope my previous reply wasn't mean to you, as I wasn't. Just got really frustrated with the support staffs of my hosting.

1)
First of all, I really like my hosting, and switching to a different one isn't my best option. Therefore I am trying to prove my hosting is right about my 100 mbps line.

Just that the real-life result is showing the other way.

2)

(01-21-2010, 03:55 AM)Terrorkarotte Wrote:  so 100000000bit:
.
.
.

The bit conversion thing, I just go by this reference (http://physics.nist.gov/cuu/Units/binary.html). I am not a tech savvy like you anyway, and that's why I am seeking help now all over the net regarding this issue (http://www.dslreports.com/forum/r23669852-Hi-need-some-help-on-verifying-my-uplink-speed).

How about we just keep with the 1/8 convention for now, as all my speed measurements are for the estimated value anyway.

3)

(01-21-2010, 03:55 AM)Terrorkarotte Wrote:  I reach exactly the same with my connections so its seems the effictive maximum for a 100Mbit line

Are you referring to this 500 MB test file? (http://limestonenetworks.com/test500.zip)

(By the way, big thanks for doing such test download for me. Very appreciate it)

4)
Continue from #3, mind to give me some results from the following sources? (Again, not to waste your bandwidth, a partial download, like 10%, will do)

- http://limestonenetworks.com/test500.zip
- http://mirror.anl.gov/pub/ubuntu-iso/CDs/9.10/ubuntu-9.10-server-amd64.iso
- http://www.gtlib.gatech.edu/pub/ubuntu-releases/9.10/ubuntu-9.10-desktop-amd64.iso

Seeing how the other 100 mbps performs with those sources, is the easiest way to verify mine.

5)

(01-21-2010, 03:55 AM)Terrorkarotte Wrote:  Read the contract. I am sure they only offer you a 100Mbit card that is connected to the internet or so. Not you get 100Mbit internetconnection.

I upgraded my uplink speed a month ago, from 10 mbps, to 100 mbps.

6)

(01-21-2010, 03:55 AM)Terrorkarotte Wrote:  If i were you start two or more downloads from fast mirrors and see how much they reach together.

That's a very good suggestion (I got the same suggestion from another forum), and I will try that now.

---

[SOLVED]
Terrorkarotte,

A big thanks to you.

1)
I took your suggestion:

Quote:If i were you start two or more downloads from fast mirrors and see how much they reach together.

and got the following result:



- Result

2)
Now I have to apologize to the support staffs for my previous arguments.

(Yet as a specialists, shouldn't they make such suggestion in the first place? Or our Terrorkarotte is beyond them. )