Welcome Guest ( Log In | Register )

Bump Topic Topic Closed RSS Feed

Outline · [ Standard ] · Linear+

 The "extreme Screamyx Slowdowns" V4, Edisi :Extreme International Slowdowns

views
     
chew13
post Jun 17 2009, 09:15 PM

New Member
*
Newbie
0 posts

Joined: Jan 2005


Reporting from JB.

Tested IP range pinging google.com
124.xx.xx.xx
60.51.xx.xx
60.53.xx.xx
60.54.xx.xx

All IPs suck with up to 40% lost packets.
Reseting router and modem also no help.

1st time problem appeared briefly at around 3-4am, then recovered about an hour.
2nd time problem started at around 2-3pm and recovered in a couple of hours.
3rd time problem stared at around 8pm and now seems to be getting worse by the minute.

Sigh....
shakehead.gif
chew13
post Jun 23 2009, 04:04 PM

New Member
*
Newbie
0 posts

Joined: Jan 2005


Reporting from JB.

Since the upgrade that should had been concluded by 20th June 2009 6am according to their announcement, I had been suffering from frequent drop packets.

- Trying to sync a database for the past 3 days to a server in US without success as the connection always drop halfway due to packet losses.
- Playing Runes of Magic on US server causes my character to die multiple lag of death.
- Pinging google.com shows at packet losses from 20%-60% at any time.

By using WinMTR to trace and ping each packet path over 200 times, we get a clear conclusion that packet losses are directly the result of Streamyx and not others.

------------------------------------------------------------------------------------------
WinMTR statistics
Host - % Sent Recv Best Avrg Wrst Last
------------------------------------------------------------------------------------
OSIRIS - 0 292 292 0 8 121 10
192.168.2.1 - 0 292 292 0 60 351 60
No response from host - 100 292 0 0 0 0 0
219.93.218.29 - 1 292 290 0 80 581 10
219.94.12.93 - 1 291 289 10 70 581 110
203.106.4.41 - 2 291 287 10 71 571 111
203.106.253.225 - 2 291 288 20 90 531 110
210.187.129.131 - 3 291 283 10 94 541 50
58.27.105.105 - 3 291 285 20 85 581 60
10.55.32.73 - 3 291 283 10 84 510 60
10.55.32.110 - 5 291 278 10 84 491 60
58.27.106.198 - 3 291 284 40 107 531 90
219.94.9.10 - 59 291 121 80 148 571 200
209.85.241.58 - 59 291 121 40 128 1142 100
209.85.249.236 - 57 291 126 100 173 1051 101
209.85.248.131 - 56 291 130 180 247 531 190
209.85.248.128 - 53 291 137 210 292 881 290
209.85.249.18 - 55 291 131 240 318 801 271
209.85.242.215 - 56 291 130 250 325 711 271
72.14.232.213 - 55 291 132 260 330 821 310
209.85.253.137 - 54 291 136 260 331 530 280
yx-in-f100.google.com - 57 291 126 250 343 791 290
____________________________________________________________________________________

1-3% loss in this test is usually normal and acceptable but anything over 3% shows where the problem lies.

We can see from the graph above that it starts from 203.106.4.41 and worsen as it gets to 210.187.129.131. Both are TM's IPs. Still minor issue here.

But when it gets to 219.94.9.10 (still is TM's IP) we see the packet loss becomes 59%. This is the road bump that TM Net should be fixing immediately.

If there are anybody from TM that are reading this, please, fix the problem. I had done some of your work for you so please do yours now so that we can all have our Broadband back. Thank you.



chew13
post Jun 25 2009, 10:34 AM

New Member
*
Newbie
0 posts

Joined: Jan 2005


Reporting from JB. On 1mbps package and currently using 60.53 ip range.

Result from trace and ping to google.com at 1 sec interval with 64kb using WinMTR over 200 times.

------------------------------------------------------------------------------------------
WinMTR statistics
Host - % Sent Recv Best Avrg Wrst Last
------------------------------------------------------------------------------------
OSIRIS - 0 231 231 0 10 151 0
192.168.2.1 - 3 231 225 0 137 17775 17775
No response from host - 100 231 0 0 0 0 0
219.93.218.21 - 4 231 224 10 81 511 120
219.94.12.89 - 3 230 224 10 72 480 110
203.106.4.41 - 3 230 224 10 72 380 160
203.106.253.225 - 3 230 224 10 89 360 221
202.188.126.132 - 4 230 221 10 100 361 171
58.27.103.97 - 4 230 223 20 82 280 150
10.55.32.73 - 5 230 219 10 89 470 180
10.55.32.118 - 7 230 215 20 85 330 150
58.27.106.238 - 32 230 158 40 107 451 161
219.94.9.10 - 54 230 107 40 105 251 151
209.85.241.58 - 44 230 130 50 110 290 100
209.85.249.236 - 52 230 111 110 169 351 110
209.85.248.131 - 53 230 110 190 251 551 280
209.85.248.128 - 48 230 121 210 278 451 230
209.85.249.18 - 44 230 130 220 297 461 291
209.85.242.215 - 44 230 129 230 326 591 330
72.14.232.213 - 49 230 118 260 473 17034 17034
209.85.253.141 - 46 230 125 270 343 561 360
yx-in-f100.google.com - 49 230 119 240 336 490 340
____________________________________________________________________________________

The result shows the bottleneck starts at 10.55.32.73 and more significantly at 58.27.106.238 (TM's IP) and worsen at 219.94.9.10 (still is TM's IP).

It appears to me that Streamyx's main problem is that they are repeatedly shoving watermelons thru a garden-hose. Sometimes part of the watermelon gets through, sometimes it stuck.

Lousy equipments may be one cause. Oversold bandwidth could be another cause. And that could be the reason why you may get high-speed but you pages still load slowly as the lost packets had to be repeatedly resend over and over again till they get through. For protocol like TCP that's still fine because you will still get all the data eventually as long as they do not time out for too long, but if you are using UDP or other protocols that do not (or have limited) acknowledge or automatically request the losses, you are in deep shit like me even though your data may not be bandwidth intensive (MMORPGs etc) due to the lack of a minimal sustainable throughput.

There is no point blaming P2P users or others for being bandwidth hogs. The real problem is that TM's infrasture are incapable of handling what they are selling and instead of improving on the infrastruture to ensure a minimal sustainable stability, they roll out ever high speed packages (think bigger watermelons) and continue trying to push them through the same tiny waterhose. The result? Profit increases while quality of service deteriorate severly at a rapid pace.

I remembered about 5 years ago when I was using streamyx on 1mbps and it was smooth as a baby's bottom for months. No drop connections and not much problem. Now it seems like we did be lucky to get a week of smooth connection every month.

As I am a foreigner and it would not be wise for me to get involve in a state-owned monopoly like TM, is there anyone here with contacts with any officials at MCMC and can highlight this to them? The poor quality of broadband services in Malaysia is severly stunning its growth and potential.

Thank you.

This post has been edited by chew13: Jun 25 2009, 10:39 AM

Topic ClosedOptions
 

Change to:
| Lo-Fi Version
0.2066sec    0.30    7 queries    GZIP Disabled
Time is now: 14th December 2025 - 06:43 PM