Welcome Guest ( Log In | Register )

Please provide proper description when you report a post. Report button abuse will earn you an automatic warn + suspension.
 
RSS feedBump TopicReply to this topicStart new topicStart Poll

Outline · [ Standard ] · Linear+

> Unifi Speed Test, Speeds shown diff to network activity

cloudnine
post Dec 10 2010, 01:15 PM


New Member
*
Group: New Member
Posts: 4

Joined: Dec 2010
I smell something fishy.

While using the unifi spped test widget: http://www.unifi.my/unifi/index.php?option...d=75&Itemid=269 i was watching the network speeds in my menu bar. They just dont match.

This is supposed to be a business package, 10Mb up and down.

Download speed is fine, i was watching 1.0 - 1.2 MB/s as the Dial shows between 8 - 10Mb/s. All good.

Then the upload test starts, i watch 100-150Kb/s while the Dial showns nothing.
Then the network all stops, about 1/4 through the test. The Dial shows nothing.
Long pause 1min+.
Suddenly without any network traffic the Upload Dial goes up to 8-10Mb, BUT NO NETWORK TRAFFIC. NONE!

This is very fishy.

This is the test they unifi peope proudly show you after installation!

This is not good! vmad.gif (My boss already to cancel the unifi.)

Anyone else seen something like this?

Thanks!
cloudnine.




Attached image(s)
Attached Image
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
vergas
post Dec 10 2010, 03:55 PM


Regular
******
Group: Senior Member
Posts: 1,245

Joined: Oct 2004



Some similarities not not sure the conclusion

user posted image

user posted image

Uploaded with ImageShack.us
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
clawhammer
post Dec 10 2010, 04:10 PM


///M
Group Icon
Group: VIP
Posts: 8,113

Joined: Jan 2003
From: Kuala Lumpur




If you're using TMnet's tool to test, it's normally testing against their own TMnet server. This would obviously be fast but if you test against some other servers around the world, speed is screwed. I guess that explains the inconsistencies you are talking about.
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
brian12988
post Dec 10 2010, 04:19 PM


Regular
******
Group: Senior Member
Posts: 1,256

Joined: Jan 2007
From: Marehsia.


QUOTE(clawhammer @ Dec 10 2010, 04:10 PM)
If you're using TMnet's tool to test, it's normally testing against their own TMnet server. This would obviously be fast but if you test against some other servers around the world, speed is screwed. I guess that explains the inconsistencies you are talking about.
*
i think cloudnine means when he is testing against TM server, his NIC shows no activity...

but cloudnine, u have to understand that although ur NIC shows liek that, it does not mean that TM is cheating u..usually upload tests use small files to test..ur network activity only shows that ur PC has uploaded the file but it takes time for the file to reach the server and then to process and measure the speed...
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
ahpek26
post Dec 10 2010, 04:20 PM


Casual
***
Group: Junior Member
Posts: 320

Joined: Apr 2007


QUOTE(clawhammer @ Dec 10 2010, 04:10 PM)
If you're using TMnet's tool to test, it's normally testing against their own TMnet server. This would obviously be fast but if you test against some other servers around the world, speed is screwed. I guess that explains the inconsistencies you are talking about.
*
You totally missed the issue. The problem is the upload test itself is inaccurate because it stopped during testing and yet the upload meter shows it's still running. I've seen this kinda behavior before in speedtest.net and download.tm.net.my on international server wheres the meter is still running and yet my speedometer (client-side monitoring) shows zero activity.

Most speedtest only takes the average of the test, not the last speed it's stated in the meter. To get the most ACCURATE TEST, download a file, preferably from international server (Hotfile, Megaupload, Rapidshare etc) and see what your getting. My recommendation is to upload something to mediafire.com (allow multiple connections for free user, which is good) and then try downloading from them (using download manager). This will tell you the most accurate result for international server.

Why not local server? Who cares about local server. There's very little stuff hosted in Malaysia so its somewhat pointless. UNLESS you want to test for local cap speed, local latency or to make sure your line (physical wire or cable) is in good condition. If they are, then your local speed and latency should be good as well.

Lastly, don't trust them! Do you own speedtest with the method I mention above and see for yourself. It's not too difficult and or technical so anyone can do it.

This post has been edited by ahpek26: Dec 10 2010, 04:21 PM
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
cloudnine
post Dec 10 2010, 04:28 PM


New Member
*
Group: New Member
Posts: 4

Joined: Dec 2010
QUOTE(ahpek26 @ Dec 10 2010, 05:20 PM)
You totally missed the issue. The problem is the upload test itself is inaccurate because it stopped during testing and yet the upload meter shows it's still running. I've seen this kinda behavior before in speedtest.net and download.tm.net.my on international server wheres the meter is still running and yet my speedometer (client-side monitoring) shows zero activity.

Most speedtest only takes the average of the test, not the last speed it's stated in the meter. To get the most ACCURATE TEST, download a file, preferably from international server (Hotfile, Megaupload, Rapidshare etc) and see what your getting. My recommendation is to upload something to mediafire.com (allow multiple connections for free user, which is good) and then try downloading from them (using download manager). This will tell you the most accurate result for international server.

Why not local server? Who cares about local server. There's very little stuff hosted in Malaysia so its somewhat pointless. UNLESS you want to test for local cap speed, local latency or to make sure your line (physical wire or cable) is in good condition. If they are, then your local speed and latency should be good as well.

Lastly, don't trust them! Do you own speedtest with the method I mention above and see for yourself. It's not too difficult and or technical so anyone can do it.
*
Here is what i am seeing. Look at the blue d/l versus the red u/l. Also look at the data totals.

Download speeds are fine: 1.0 - 1.2 Mb/s. Uploads i see are 100-150 Kb/s.

How can the tm speed meter claim those upload speeds?


Added on December 10, 2010, 4:35 pm
QUOTE(brian12988 @ Dec 10 2010, 05:19 PM)
i think cloudnine means when he is testing against TM server, his NIC shows no activity...

but cloudnine, u have to understand that although ur NIC shows liek that, it does not mean that TM is cheating u..usually upload tests use small files to test..ur network activity only shows that ur PC has uploaded the file but it takes time for the file to reach the server and then to process and measure the speed...
*
Thanks. I think what I am finding is totally bogus is the animation of the upload dials. They bear no relation to the data uploaded, both in the size or the timing.

Just some fancy useless animation!


This post has been edited by cloudnine: Dec 10 2010, 04:35 PM


Attached thumbnail(s)
Attached Image
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
brian12988
post Dec 10 2010, 04:42 PM


Regular
******
Group: Senior Member
Posts: 1,256

Joined: Jan 2007
From: Marehsia.


QUOTE(cloudnine @ Dec 10 2010, 04:28 PM)
Here is what i am seeing. Look at the blue d/l versus the red u/l. Also look at the data totals.

Download speeds are fine: 1.0 - 1.2 Mb/s. Uploads i see are 100-150 Kb/s.

How can the tm speed meter claim those upload speeds?


Added on December 10, 2010, 4:35 pm

Thanks. I think what I am finding is totally bogus is the animation of the upload dials. They bear no relation to the data uploaded, both in the size or the timing.

Just some fancy useless animation!
*
thats because usually it is..one thing is that when measuring upload speeds, the bigger the file the better...small files tend to give a slightly off measurement..

its best u test using measurementlabs tools instead.. measurement lab..
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
wKkaY
post Dec 10 2010, 09:27 PM


stellar
Group Icon
Group: Forum Admin
Posts: 5,575

Joined: Jan 2003
From what I see, it uploads a 30KB chunk multiple times. It can be uploaded in 24ms with a 10Mbps connection, which might be too brief to get an accurate sample. Taking multiple samples as they did would help reduce the error though.

For what it's worth, I just did a upload test with my friend using Unifi in Subang - he could upload to the lowyat.net servers at 20Mbps. So I no longer doubt that you're getting what you paid for tongue.gif
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
Jeremy_John
post Dec 11 2010, 09:24 AM


Getting Started
**
Group: Junior Member
Posts: 71

Joined: Jan 2010
From: Klang


QUOTE(cloudnine @ Dec 10 2010, 04:28 PM)
Here is what i am seeing. Look at the blue d/l versus the red u/l. Also look at the data totals.

Download speeds are fine: 1.0 - 1.2 Mb/s. Uploads i see are 100-150 Kb/s.

How can the tm speed meter claim those upload speeds?


Added on December 10, 2010, 4:35 pm

Thanks. I think what I am finding is totally bogus is the animation of the upload dials. They bear no relation to the data uploaded, both in the size or the timing.

Just some fancy useless animation!
*
thats http://speedtest.streamyx.com.my right?
If yes then how the hell did you manage to test Unifi on that SpeedTest because i could never used http://speedtest.streamyx.com.my my result were always 100kbps and below for UP and DOWN worst then a streamyx 512 package doh.gif
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
akash3656
post Dec 11 2010, 10:36 PM


Getting Started
**
Group: Junior Member
Posts: 282

Joined: Feb 2009
Can't you just try the trusted www.speedtest.net ???

Use local server Cyberjaya... cos that 1 is hosted by TM.... and you seem to love TM's own fail speedometer

Should be more accurate than TM's own speedometer by any day..
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
Oga
post Dec 12 2010, 07:05 PM


New Member
*
Group: New Member
Posts: 2

Joined: Dec 2010
QUOTE(akash3656 @ Dec 11 2010, 10:36 PM)
Can't you just try the trusted www.speedtest.net  ???

Use local server Cyberjaya... cos that 1 is hosted by TM.... and you seem to love TM's own fail speedometer

Should be more accurate than TM's own speedometer by any day..
*
No offense, but what makes you think that www.speedtest.net is trusted? Just because they're the most well-known doesn't make them trusted. I say this because their speedtest always shows full speed for my connection but in reality I'm getting about 30% of the original speed since mid November 2010.

Either they are not to be trusted or TMnet has figured out a way by tampering the results shown to Streamyx/Unifi subscribers in such a way that every time you perform a speed test, you will be shown full speed.

Cheers.
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
brian12988
post Dec 13 2010, 11:46 AM


Regular
******
Group: Senior Member
Posts: 1,256

Joined: Jan 2007
From: Marehsia.


QUOTE(Oga @ Dec 12 2010, 07:05 PM)
No offense, but what makes you think that www.speedtest.net is trusted? Just because they're the most well-known doesn't make them trusted. I say this because their speedtest always shows full speed for my connection but in reality I'm getting about 30% of the original speed since mid November 2010.

Either they are not to be trusted or TMnet has figured out a way by tampering the results shown to Streamyx/Unifi subscribers in such a way that every time you perform a speed test, you will be shown full speed.

Cheers.
*
err...pretty sure it is not like that....speedtest.net is only to be used as a basic measurement only....as it does not take into other factors...
if u really want to test, use measurement labs tools...
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
andrew9292
post Dec 13 2010, 12:37 PM


-/Livin' On A Prayer/-
*****
Group: Senior Member
Posts: 867

Joined: Sep 2008
From: Petaling Jaya


QUOTE(Oga @ Dec 12 2010, 07:05 PM)
No offense, but what makes you think that www.speedtest.net is trusted? Just because they're the most well-known doesn't make them trusted. I say this because their speedtest always shows full speed for my connection but in reality I'm getting about 30% of the original speed since mid November 2010.

Either they are not to be trusted or TMnet has figured out a way by tampering the results shown to Streamyx/Unifi subscribers in such a way that every time you perform a speed test, you will be shown full speed.

Cheers.
*
i guess it's normal... been experiencing it since my first 512k line when streamyx just launch. I guess they just take the average of the readings. But i notice something weird tho. If i'm on wireless (wifi) it happens... but if i'm on wired (cable to router).. the NIC shows activity all the way till the end of the test. But the figures on speedtest/ any speedtest system is mostly the average... eg if 10sec u are at 2M then the next 10 sec ur speed drops to 1M it will show 1.5M at the end of the test

This post has been edited by andrew9292: Dec 13 2010, 12:38 PM
User is offlineProfile CardPM
Go to the top of the page
+Quote Post
darrenlim89
post Feb 15 2014, 12:29 PM


New Member
*
Group: Junior Member
Posts: 12

Joined: Sep 2006
user posted image

There is 0.21mbps less than the desired speed. Is that normal?
User is offlineProfile CardPM
Go to the top of the page
+Quote Post

Bump TopicReply to this topicTopic OptionsStart new topic
 

Switch to:
| Lo-Fi Version
0.0429sec    1.53    6 queries    GZIP Disabled
Time is now: 19th April 2014 - 07:20 AM