Calling all IS ADSL users.

[Ipwn 4]

Hi guys,

Over the past few weeks I have noticed that ISPs on the IS backbone seem to be doing very very well at speed tests using either MyBB or Speed test . net. The problem comes in when one makes use of other servers to do these tests again:

On a WA uncapped account I was pushed to the highest shaping level after just 30GB (ridiculous but this is a topic for another discussion) at which point is was still getting line speed on all speed tests. The moment I connect to anything else however I have less then dial up speeds. I performed some tests into our corporate network to ensure that I can monitor capacity on both ends to prove the account is being slowed down and confirmed that I was shaped.

This to me is slightly unethical, if you are shaping your accounts to death show it on the speed tests, don't try and make your stats appear better.

What's your experience been like? This behaviour occurred on WA and OW accounts for me.

 

[Keegan]

If I'm not mistaken, speedtest's use port 80, which is normal http traffic.

Thus websites will load as usual when throttled/shaped compared to when your are not.

Once you start downloading etc, then that traffic becomes throttled...

 

[wingnut771]

Hi guys,

Over the past few weeks I have noticed that ISPs on the IS backbone seem to be doing very very well at speed tests using either MyBB or Speed test . net. The problem comes in when one makes use of other servers to do these tests again:

On a WA uncapped account I was pushed to the highest shaping level after just 30GB (ridiculous but this is a topic for another discussion) at which point is was still getting line speed on all speed tests. The moment I connect to anything else however I have less then dial up speeds. I performed some tests into our corporate network to ensure that I can monitor capacity on both ends to prove the account is being slowed down and confirmed that I was shaped.

This to me is slightly unethical, if you are shaping your accounts to death show it on the speed tests, don't try and make your stats appear better.

What's your experience been like? This behaviour occurred on WA and OW accounts for me.

I've noticed when there's shaping, I would get full speed on speedtest, but only because it's local, if you tried testing to international (washington dc speedtest server is my favourite as it always maxes the line when there's no shaping (after 5pm)), you would see a slowdown on performance.

P2P/NNTP is normally shaped more than HTTP during these times to.

Best to adjust your usage patterns and make use of the 12-6am window to get the most out of IS.

 

[Ipwn 4]

I've noticed when there's shaping, I would get full speed on speedtest, but only because it's local, if you tried testing to international (washington dc speedtest server is my favourite as it always maxes the line when there's no shaping (after 5pm)), you would see a slowdown on performance.

P2P/NNTP is normally shaped more than HTTP during these times to.

Best to adjust your usage patterns and make use of the 12-6am window to get the most out of IS.

Nope, I use local servers (non speed test servers) which are used to monitor our client's fibre networks to test against. The actual tests consists of and http transfer (1 large and 1 which downloads 50 odd smaller files) and then encrypted downloads. The ISP can see where the traffic is coming from but not what is inside.

point is these tests all tell me I am shaped, when I do a speedtest (which is the best know way to check your internet) everything is fine.

 

[Ipwn 4]

If I'm not mistaken, speedtest's use port 80, which is normal http traffic.

Thus websites will load as usual when throttled/shaped compared to when your are not.

Once you start downloading etc, then that traffic becomes throttled...

The port has nothing to do with it, any traffic that is not encrypted can be intercepted and inspected. I can very easily recreate shaping on our firewalls which can give me a 10Mbps speedtest but limit the total throghput of all other traffic to half of that.

One portocal which should never be shaped is RDP and that is also slow...

 

[stephannel77]

The port has nothing to do with it, any traffic that is not encrypted can be intercepted and inspected. I can very easily recreate shaping on our firewalls which can give me a 10Mbps speedtest but limit the total throghput of all other traffic to half of that.

One portocal which should never be shaped is RDP and that is also slow...

Ditto, and some of the shapers automatically picks up ookla speed tests by the way the packets are inspected. We also host an ookla speedtest platform at MTN and shaping is set to highest priority. So it is with every ISP out there. There is a tread on myBB where you can dl a diagtool that will give you more accurate results.

 

[Ipwn 4]

Ditto, and some of the shapers automatically picks up ookla speed tests by the way the packets are inspected. We also host an ookla speedtest platform at MTN and shaping is set to highest priority. So it is with every ISP out there. There is a tread on myBB where you can dl a diagtool that will give you more accurate results.

Thanks for confirming.

The issue through is that IS unshapes these tests making their network appear to perform better then others who dont try and hide shaping.

 

[stephannel77]

Agreed. Even though it is a shaped service they offer. it should at least perform at half the capacity of your ordered service and not less. I would really like to know what shapers they use? I know mweb uses Allot communication though.