Changes to 256 Package

ShadowSA

Well-Known Member
Joined
Apr 3, 2004
Messages
250
Reaction score
0
Location
South Africa.
Hi all,
I had an open speed ticket for a while now, so some dude phoned me and said they made some changes to the 256 package and apparently it's amazing. Can anyone on a 256 confirm this? I'm on a 128 and the ETA for the same changes to be made on the 128 is end of Jan 2005.

Local speeds are fine, International I get 4Kb/sec FLAT, no more no less.
I'm on Helderkruin and have been disconnected more than a few times this weekend.

Cheers.
 
I don't have an answer for you ShadowSA.

I just want to ask all the people that can't give constructive comments/suggestions regarding Sentech rather to not post at all. I come here to read about mywireless not about how canceling mywireless is the best thing you've ever done.
 
The last couple of weeks my international speed tests have all been averaging around 200kbps on the 256k package, so (touch wood) it does seem to have improved. Guess I jinxed it now by admitting it. ;)

I'll be able to get the link tomorrow though (running the tcpIQ automated test every three hours) - Sentech is set up at work now and not at home. Will check and post again.
 
I can confirm I'm getting about 25KB/s on my 256K from an international ftp....for the first 400K, then it drops to 8KB/s flat :(

Local is flying as usual :D
 
Getting the same speeds here. Full speed for the first 500K - 1MB. Then it gets capped at 8K/sec.

Seems like a ploy to trick speed tests. Most of the tests I've tried are around 500K.

Local download speeds are fast
 
I am also getting the same as passif, nonroker and JayT. Downloads start at ±25kB but drop to 8kB after about 600k. If you are downloading a file that is cached on their proxy, it doesn't seem to get throttled - last week one of my users downloaded XP SP2 and the whole 272MB file came down at ±25kB, taking about 3 hours to download.
 
Anyone else heard about the planned international bandwidth upgrades in Jan 2005?
 
nahh... I heard they were first going to fix the 128K's then work through the 256 and 512. So it seems they are now fininshed with 128K's and 256K's.

Still think its better then it was but not better then it was right in the beggening.

ShadowsSA I dunno why they said they still have to work on the 128K's cuz as far as I know they are done.
 
Swindletech has done it again with their little tweaks, 256 downloads at full speed for the for the first segment, then caps at 9/10, 128k also, but caps at 4/6.
All this tinkering is doing is fooling bandwidth tests.
And I've just about had it.

After a speed ticket being open for several months, I get a call from one of those horrid orcs 2 weeks ago, asking me rather excitedly if things are fixed, I tell him I'm not sure yet, as I haven't been home that day.
He then goes on to tell me he's closing the speed ticket with or without my consent, and they are closing off every other speed ticket, as far as they concerned, there is no more they willing to do!

Needless to say, we know how the speed looks, I just haven't had the patience or energy to call back and log another ticket!
actually, I've just had enough and done it now, apparently another one of you guys did the same this morning.
 
That has to be the sneakiest business decision ever. Damn - how I'd loved to get hold of that meeting's minutes.
So - Use a download manager and configure it to download in segments not exceeding 500 kb. That'll get them, and give you the advertised download speeds.
 
JayT said:
Getting the same speeds here. Full speed for the first 500K - 1MB. Then it gets capped at 8K/sec.

Seems like a ploy to trick speed tests. Most of the tests I've tried are around 500K.

Local download speeds are fast


Do an Int test here http://performance.toast.net/ the f16 jets is over a meg and gives me an average 81k on my "128k package" - on my laptop, at the front of the house where signal is stronger, on my desktop, back of the house - only 49k
go figure.
 
noswal, thanks for the link. Here's my results. Clearly showing the slow down the moment the download is over 500K:

Text (341K) - Lunar Pages (result: Your throughput: 217* K )
Shuttle (414K) - Lunar Pages (result: Your throughput: 213 K)

Shuttle + Text (755K) - Lunar Pages (result: Your throughput: 143* K )
Mars Rover (878K) - Lunar Pages (result: Your throughput: 119* K )

F-16 Jets (1,397K) Lunar Pages (result: Your throughput: 87* K )
 
yeah I just got a mail from Sentech asking for speed tests.

I used the above link so i got a lousy 46Kb when using the Rover speed test but hey dont need to know that :P

I told them that as far as I concerned the percieved speed may be quicker but in reality still crappy.
 
Here is an opportunity for someone to write a small piece of software that stops downloading at 400kb and then starts again. That way, we should average some good download speeds. I will be willing to pay for this.
 
Here is an opportunity for someone to write a small piece of software that stops downloading at 400kb and then starts again.
I got something I wrote recently for linux to abuse the multithreading capabilities of the connection, unfortunately, It works on the application layer, so I have so far only got it working on the http protocol, and they moved over to the new bandwidth profiler a week be4 I completed it (but nonetheless, it was a good holiday project to take on).

If things are still the same when I return from my road trip ( I leave tomorrow, and return mid january) I'll add some functionality to change it fom multiple threads to a single thread, but make the thread limit 400k, with a little sleep before it opens the next thread segment.

before I go ahead with it, it will require a little testing as to how long after a 400k segment is downloaded, the speed will return, it might be an exercise in futility if the timeout be4 max speed is too big.

On a 128K, if 2minutesof sleep is required, this is equivalent to the time required to download 400k, so this will prove to be a pointless idea.

well, lets see how it looks in the new year!
If things are the same, and that timeout is relatively small, I'll modify it slightly, or put in some switches for the mode bit.

I will be willing to pay for this.
I'll be willing to provide it free just to fsck Sentech on their cheap attempt at making their network look good.
heck, I'll even attempt porting it to windoze to ensure more people screw sentech.
 
Last edited:
Does their bandwidth manager not work per TCP/IP session? If so you would basically need to create a new session in order to get the burst speeds. In this scenario you could try creating two threads which flip-flop i.e. thread 1 creates a session and starts downloading. As thread 1 approaches 400k you fire up thread 2 to create a session starting at 400k. When thread 2 approaches 800k you fire up thread 1 again starting at 800k (hope this makes sense ;))

You could use more than 2 threads however it probably won't make any difference. The idea of the flip-flop is to eliminate the session creation delay. Unfortunately I don't have MW to try it out.
 
Sentech's new "trick" doesn't work that way, your IP gets limited to 8KB/s (for 256K) intl speeds, multiple threads don't help. Once a download is capped at 8KB/s, all other intl downloads/browsing becomes slow.
 
I can also confirm the 256k package download behaviour - first 500k full speed then slows down to a crawl.

A download manager that would automatically stop and restart every 500k would be brilliant - this works if you do it manually, i.e. when manually stopping and resuming the download every 500k it keeps on reverting to full speed again. It's somewhat tedious doing this for any download greater than a few megabytes, though, so a dl manager that does this automatically will work brilliantly.

I suspect, though, that if we find a download manager that does this then Sentech will change the way downloads work just as quickly, because, as we all know, they are hellbent on providing the crappiest service possible.
 
Top
Sign up to the MyBroadband newsletter
X