Now getting 1.23KBps on my 256k - AGAIN!!!

[SINTech]

LETTER TO SENTECH AFTER SPEED TEST

Speed Test
Your download speed is: 1.23 KBps, or 9.87 Kbps
The test took 829.922 seconds to complete
To whom it may concern

I’m really getting frustrated with my 256k connection at this stage. It’s getting worse each week and this is not the service as agreed upon in the contract. I gave Sentech the last few times 3 to 4 days to resolve this kind of issues but this is becoming totally ridiculous as well as extremely bad service and standards that Sentech is maintaining. If this is not resolved and the ticket is not closed by myself by tomorrow COB, this letter as well as the previous letters will be email to the MD. If this fails to have a impact I’ll come in person to your premises to resolve this issue with supported documents.

Please also find below an email sent to me from a stranger(Gregory Hinton Nietsky) which obviously found my details on YOUR database which is also a non disclosure bridge according to the contract.

When is all this going to stop and when will your guys start to get their ducks in a row? This is much worse than Telkom could ever dream of and it’s totally unbelievable that I’m still paying for this service while Sentech doesn’t even try to reimburse any of my fees paid.

Then Sentech has the liberty to send me an email regarding download “abuse” or “misuse”. On a “broad band” line like this, your email is totally irrelevant.

Also note that I’ll email this document as well as the rest of them to IT Web if this is not resolved by COB tomorrow. However, I’ll publish this email on the myadsl website regardless.

It seems your broadband is measured at the speeds as per when the 9600 modems just came out. Who in this world is getting 1.23KBps and calls it broadband while the speed test is taking me 13 minutes to complete?

What is your LONG-TERM plan to resolve this issue once again? I don’t want your short-term plan please!!!

My speed test results are as follows:
Speed Test
Your download speed is: 1.23 KBps, or 9.87 Kbps
The test took 829.922 seconds to complete


I think if Sentech is not going to shape-up, they’ll ship out

 

[Donovan]

<blockquote id="quote"><font size="1" face="Verdana, Arial, Helvetica" id="quote">quote:<hr height="1" noshade id="quote"> found my details on YOUR database which is also a non disclosure <b>bridge</b> according to the contract<hr height="1" noshade id="quote"></blockquote id="quote"></font id="quote">

Breach.

Sorry []

 

[donn]

I assume you got a job card number? If not, get one anyway. That is, if you still have a phone line []

I know this seems like a pointless exercise, but at least you are forcing them to log the call. If you keep logging a new call every week eventually you will have a case to either cancel the contract or downgrade it to 128k.

If you don't get a response, phone an log a new complaint, and insist that they note all the other job card numbers in that job card too.

The weird thing is that I just downloaded http://www.sentech.co.za/speedtest/512k.gif at 9KB/s from a 128k dedicated line connected through Datapro. So it seems the outside world can get it faster than a WM user.

Download took from 11:07:43 to 11:08:49 for 534.73k
Donn Edwards
http://privacy.4mg.com

 

[SINTech]

Thanks for the correction Donovan.

Anyways, I did insist on a ticket and they did give it to me so from that point of view it's fine.

 

[Jacques]

Just for fun I also downloaded the file:
www.sentech.co.za/speedtest/512k.gif at an average speed of 4.6KB/s using Mweb 56K dial-up connection.

...wonder if I might in fact have a broadband connection!? []

Jacques
CT

 

[louisp]

Love your post Jacques!!

Puts things in a very clear perspective!!

MWEB should upgrade their advertising - MWEB 56K better than broadband !!

 

[Jacques]

must be the "and all and all" I'm paying for! []

In reply to a question about speed, Sentech-support reminded me that it was a contended service, and that speed could go "as low as 4.2kbps"! [:0]
Just wondering how contended your average ISP's dial-up service is. [?]

Jacques
CT

 

[loosecannon]

FFS that is attrotious and not the "fast internet" they advertise comments from helldesk like this are official sentech statments and this is surly good for the ASA case....

that is 3,2% of 128k when i was running a ISP in the early days we only bothered with contention on international as local was cheap [if you call telkom cheap] in otherwords we tried to ensure all nodes never average above 75% utilization at any given time and that the core links were a % of network load at all times so international would be 15-25% of load to maintain the ratios things like TX proxing were employed to save this resource ... customers were encourage to use south african resources as MUCH AS THEY WANTED and to expect speed problems on international links ... granted we did not have P2P in those days but i guess i would do QOS/TOS control if i was to be in the same position ...

quite frankly the same problem happened at sentech as at telkom the ape setting up the buissiness case did not have a clue and when i raised the issue with the ape concerned at telkom pre release it was like headbutting a brick wall i have sat in management meetings when capping has been discused i was amazed at some of the proposals [far worse than current reality]

i am currently getting max of 2500 B/s

http://firewall.networksentry.co.za/mrtg/

 

[mrbob]

When the sentech salesman told me about the contention ratio at the Sandton Computer Fair this year he
said it was something that would only occur very rarely, he said at the very least one should expect about 10Kb/s (on 128k package). The point is that sentech calls this a 128k package , I am yet to see download speeds of 128k , **** it.. I almost wet my pants when my speeds average 64k (which is so rare it's sad) , everyone knows about the contention ratio.. but they cannot keep using it as an excuse.

Rabble rabble, bitch bitch ARRRGH

 

[donn]

What gets me about this whole contention thing is that it should only be a problem

a) If your particular tower is "overloaded"
and/or
b) If you're trying to connect to a service outside the Sentech network

If you take 2000 customers and divide it by (30 x 3 nodes/tower) you get 22 towers. But as far as I know there are only 32 in service. What's more, if more than 90 users attempt to log in to a given tower the 91st user gets told that there aren't any ports. Aparrently this hasn't happened yet, so none of us should be getting bad bandwidth, in theory.

It's all very confusing. At the same time the modem box talks about download speeds of 3 MB, so does that mean each tower is theoretically capable of 3x30= 90MB, in which case the whole theory goes bang. I just wish they would explain the system properly to users before they signed up.

Donn Edwards
http://privacy.4mg.com

 

[chuckl]

Donn, As I understand it, and please feel free to correct:

Theoretical max speed the modem 'could' transfer is 3Mb, in a perfect world, probably lab conditions.
In the towers we have NodeB's which have the backhaul links which theoretically go to the internet. In practice they meander round through infosat and routers for a while. On a tracert it is 6 hops to www.sentechsa - go figure.
These backhaul links comein various flavours, depending on the NodeB employed. Anything from 6Mb/s entry level to 36Mb/s full out - do the arithmetic.
I also suspect it is 30 users per time slot per antenna, but have no proof of this. IPWireless also talk about 6 way combiners, which would imply to me that multiple NodeB's can use the same antennae.
Perhaps Rodent or someone in the rf game can clarify

 

[Ditch]

<blockquote id="quote"><font size="1" face="Verdana, Arial, Helvetica" id="quote">quote:<hr height="1" noshade id="quote"><i>Originally posted by donn</i>
<br />
If you take 2000 customers and divide it by (30 x 3 nodes/tower) you get 22 towers. But as far as I know there are only 32 in service.
<hr height="1" noshade id="quote"></blockquote id="quote"></font id="quote">
Seems likely Sentech are lying when they claim (continually) that the low speed is the effect of the contention ratio. For anyone to consistently get speeds of 2KB/s or less, at a contention ratio of 30:1, at least 8 out of every 30 users would have to be downloading full blast for long periods of time. That's 27% of users, or about 540 users out of Sentech's +/- 2000 users. Now that itself is incredibly unlikely on the face of it, but more importantly, Sentech themselves state that only a "small percentage" of users are "abusing" the service, in fact according to them no more than 6%. So let's be generous and say that only 4 out of 30 users are downloading at any time. That's 13.5% of users, and STILL more than double the 6% figure quoted by Winston Smith. Let's be even more generous and say that only 2 out of 30 users are downloading at any time (which actually sounds reasonable), that is 6.7% and is *still* more than the "6% of users are abusers" that Sentech claims. So the actual number of "abusers" (6%, 2 out of 30) is in reality well within norms, in fact I would have expected it to be more, so why is there so little BW? If only two or three users download full blast all the time, how can people consistently get 2KB/s or less for long periods of time? The only possible explanation seems to be that Sentech simply either doesn't have the bandwidth, or doesn't know how to run their network, this surely can't be the 30:1 contention ratio.

More 'evidence' that we are not seeing 30:1 contention ratio problems is quite simply that 30:1 is not an unusually high contention ratio by ISP standards, and yet all other ISPs in the world manage to do just fine at that sort of contention ratio. I highly doubt that Sentech users are statistically many times more likely to be downloading at any given time than any other ISP in the world.

According to them, and if we can believe this: "6% of users are using more than 50% of the bandwidth". So maybe we can make a "rough calculation" of how much bandwidth they have .. if 6% of users use 50% of BW, contended 2:1 (as 6% is roughly 2 out of 30 users), then that's 120 users getting 128/4=32kbps continually. 120x32=3840kbps=3.8Mbps. Add the 'other less than 50%' BW usage and you can guess they may have about 8Mbps. To provide 2000 users with 128kbps truly shared 30:1 (assuming all are 128 users), you need at least 8.5Mbps. It seems to be pretty close though to what they need ... so there should very rarely be contention problems. But such rough guesses may be way off.

My current speed: 1.8KB/s. Generally these days it seems to hover between 3KB/s and 5KB/s, which I consider acceptable only if I lower my standards and expectations.

 

[donn]

I just did another test at home, using a 128k MW link:

http://www.sentech.co.za/speedtest/512k.gif
Started off around 12KB/s, wandered all the way down to just about nothing. Averaged 3.51, or around 50% of a 56k modem.

Tried it again, using 8 simultaneous download threads:
It was averaging about 12KB/sec until it got halfway, then nothing. After tweaking the download manager a bit, I got an average of 8.09! That's better than 64k! I'm in shock.

My only explanation is the www.sentech.co.za server is as broken as the rest of the network.

Theoretically if 90 users with 512k contracts all connected at the same time they would get 512k/30 = 17k, i.e. 1.9KB/sec (i.m dividing by 9 to allow for a stop bit) So in the worst case noone should ever get below this because the tower would have to be able to do 512k/30*90 to keep up with demand.

So theoretically 1.23KB/sec is impossible, even under the worst case scenario, unless perhaps all 90 users were doing the speed test at the same time.

I am going to point out to the ASA that there is no attempt anywhere on the Sentech or NoWire web sites to explain what "shared 128Kbps" is or means, or how it works. In fact, NoWire gives the impression that a 256k user will never get speeds lower than 128k. Weird. It sounds a bit like MWeb saying "3x faster" but failing to justify how they get that number.

Donn Edwards
http://privacy.4mg.com

 

[dorris]

hmmm, interesting
Theory 1:
notice the date , anyone got logs for the last 3 months, is there a trend of usage going right down at the end of the month??? maybe Rat didn't find all their holes.
Theory 2:
Sentech forgot to get the latest patch from their highly advanced Norton Antivirus professional, and have thir network overloaded by MyDoom.
If even one of their PC's with no bandwidth limitations logged onto the backbone (ie Marcelles)
and got infected, he could be causing untold damage to all our presciousssssssss bandwidth

Alternatively, Sentech are just a blithering bunch of Buffoons.
and maybe its just the call centre downloading backdoor Sluts 1 to 86
Yeah , I know, call centre aint on the backbone, get a sense of humour before flaming me.

<center>Log your info at MyWireless Survey</center>

 

[donn]

Another observation: at work I connect via a 128k dedicated line to Datapro. Even during the day when there are other users surfing and so on, I can go to news.bbc.co.uk and listen to a 2 hour recording of BBC Newshour that is recorded at 32k. The RealPlayer download graph (Tools-&gt;Playback Statistics-&gt;Bandwidth) is quite smooth, with the occasional dip. Even if it averages down to 16k or even 8k, the line is smooth.

I just tried the same thing from home, amd the graph is all over the place like a yo-yo. It eventually averaged out at 8k. If I look at the Tx and Rx lights on the modem there are long pauses when nothing happens, the kind of thing you get on a dial-up landline when the quality is shot. So what is happening?

I observed the same thing when doing the 512k.gif test. My gut feeling is that if a virus was doing the rounds we would see a lot of modem activity when we aren't doing anything, rather than a lot of modem INACTIVITY when we're trying to do something.

Take a look at the tracert and ping results. Can anyone really tell me that a wireless network could be this bad?


<font size="1">Tracing route to www.sentech.co.za [66.18.65.115]
over a maximum of 30 hops:

1 96 ms 139 ms 119 ms
2 197 ms 121 ms 118 ms
3 195 ms 140 ms 119 ms gige-0-0-102.rtr-core5-stp.infosat.net [66.18.65.106]
4 * 99 ms 119 ms gige-0-0-6.rtr-core4-stp.infosat.net [66.18.67.254]
5 115 ms 121 ms * pf-hosting1b-stp.infosat.net [66.18.67.203]
6 157 ms 120 ms * www.sentech.co.za [66.18.65.115]
7 158 ms 139 ms 159 ms www.sentech.co.za [66.18.65.115]

Trace complete.

Pinging www.sentech.co.za [66.18.65.115] with 32 bytes of data:

Reply from 66.18.65.115: bytes=32 time=117ms TTL=251
Reply from 66.18.65.115: bytes=32 time=93ms TTL=251
Reply from 66.18.65.115: bytes=32 time=91ms TTL=251
Request timed out.
Reply from 66.18.65.115: bytes=32 time=83ms TTL=251
Reply from 66.18.65.115: bytes=32 time=179ms TTL=251
Reply from 66.18.65.115: bytes=32 time=98ms TTL=251
Reply from 66.18.65.115: bytes=32 time=93ms TTL=251
Request timed out.
Reply from 66.18.65.115: bytes=32 time=115ms TTL=251
Reply from 66.18.65.115: bytes=32 time=90ms TTL=251
Request timed out.
Reply from 66.18.65.115: bytes=32 time=91ms TTL=251
Reply from 66.18.65.115: bytes=32 time=111ms TTL=251
Reply from 66.18.65.115: bytes=32 time=86ms TTL=251
Reply from 66.18.65.115: bytes=32 time=127ms TTL=251
Reply from 66.18.65.115: bytes=32 time=102ms TTL=251
Reply from 66.18.65.115: bytes=32 time=222ms TTL=251
Reply from 66.18.65.115: bytes=32 time=116ms TTL=251
Reply from 66.18.65.115: bytes=32 time=97ms TTL=251
Reply from 66.18.65.115: bytes=32 time=92ms TTL=251
Reply from 66.18.65.115: bytes=32 time=108ms TTL=251
Reply from 66.18.65.115: bytes=32 time=122ms TTL=251
Reply from 66.18.65.115: bytes=32 time=137ms TTL=251
Reply from 66.18.65.115: bytes=32 time=112ms TTL=251
Reply from 66.18.65.115: bytes=32 time=88ms TTL=251
Reply from 66.18.65.115: bytes=32 time=103ms TTL=251
Reply from 66.18.65.115: bytes=32 time=118ms TTL=251
Reply from 66.18.65.115: bytes=32 time=113ms TTL=251
Reply from 66.18.65.115: bytes=32 time=74ms TTL=251
Reply from 66.18.65.115: bytes=32 time=130ms TTL=251
Reply from 66.18.65.115: bytes=32 time=85ms TTL=251
Reply from 66.18.65.115: bytes=32 time=110ms TTL=251
Reply from 66.18.65.115: bytes=32 time=173ms TTL=251
Reply from 66.18.65.115: bytes=32 time=112ms TTL=251
Reply from 66.18.65.115: bytes=32 time=371ms TTL=251
Reply from 66.18.65.115: bytes=32 time=108ms TTL=251
Reply from 66.18.65.115: bytes=32 time=107ms TTL=251
Reply from 66.18.65.115: bytes=32 time=103ms TTL=251
Reply from 66.18.65.115: bytes=32 time=96ms TTL=251
Reply from 66.18.65.115: bytes=32 time=93ms TTL=251
Reply from 66.18.65.115: bytes=32 time=87ms TTL=251
Reply from 66.18.65.115: bytes=32 time=102ms TTL=251
Reply from 66.18.65.115: bytes=32 time=396ms TTL=251
Reply from 66.18.65.115: bytes=32 time=92ms TTL=251
Reply from 66.18.65.115: bytes=32 time=86ms TTL=251
Reply from 66.18.65.115: bytes=32 time=841ms TTL=251
Reply from 66.18.65.115: bytes=32 time=116ms TTL=251
Reply from 66.18.65.115: bytes=32 time=93ms TTL=251
Reply from 66.18.65.115: bytes=32 time=85ms TTL=251
Reply from 66.18.65.115: bytes=32 time=122ms TTL=251
Reply from 66.18.65.115: bytes=32 time=97ms TTL=251
Reply from 66.18.65.115: bytes=32 time=353ms TTL=251
Reply from 66.18.65.115: bytes=32 time=107ms TTL=251
</font id="size1">
Ping statistics for 66.18.65.115:
Packets: Sent = 54, Received = 51, Lost = 3 (5% loss),
Approximate round trip times in milli-seconds:
Minimum = 74ms, Maximum = 841ms, Average = 130ms

Can someone who has a better idea please explain this to me? I am just a humble database programmer.

Donn Edwards
http://privacy.4mg.com

 

[dorris]

That looks about right, in fact, most people would be over the moon to see such good response (from myWi), I personally see slightly better, my avg time sits around 110, but 've been one of the lucky ones!
in regards to packet loss, 3/60 is also considered good (in terms of MyWi)

<center>Log your info at MyWireless Survey</center>

 

[donn]

I have a graphic of the BBC News Live feed
http://privacy.4mg.com/BBCNewsLive.gif

Also, some more detailed Ping stats
ping -t -w 20000 www.sentech.co.za
That means that a packet will not be "lost" for 20 seconds:
Ping statistics for 66.18.65.115:
Packets: Sent = 166, Received = 161, Lost = 5 (3% loss),
Approximate round trip times in milli-seconds:
Minimum = 64ms, Maximum = 451ms, Average = 109ms

Still, there is a 3% loss. So is there a link somewhere that is broken?

Donn Edwards
http://privacy.4mg.com

 

[groenie]

I logged a ticket this morning just to make sure they don't think the speed problem has gone away. I get 30kbps on local speed tests and 0 (yes that is zero) international bandwidth. Since my local bw is more than 4.6kbps, I didn't qualify for a ticket, but luckily the complete lack of international bw makes me qualify.

The only international web sie I can see is google, and that is after waiting for almost a minute - this is worse than capped ADSL.
Strangely enough, it seems HTTP traffic is worse than FTP and P2P. I can ftp from ftp.cdrom.com at 2.2kBps.

 

[TheRoDent]

Today, I received feedback from the helpdesk from my outstanding speed ticket over over 2 months.

Firmware upgrades are taking place, on advice from their international partners, and they are implementing a new hardware-based bandwidth management solution to ensure that "baddie" people don't make the experience a bad one for everyone else.

Finally, a step, and and admission of action in the right direction.


<center><h6> MyWireless <s>Hacks</s> Tweaks & Tech Info || Have you checked the fawking FAQ? <br /> <font color="red">Tired of Sentech's bad service? Want to compare speeds? We at least listen...</font id="red"></h6></center>

 

[gripen]

I was told exactly the same thing. Also that previously and currently (loose quoting here) each IP address is being allocated bandwidth per connection. Thus, people with many connections are hogging the bandwidth (for those people confused, a connection , in this sense, is a link from your PC to another PC over which data travels ie. with a website). So somehow with the hardware and software upgrades that take 3 weeks to install this is going to improve. It was a call centre orc so I dont know if quite knows what he is talking about.

Maybe what he was trying to say was the Sentech was on purpose cycling the bandwidth amongst a revolving group of IP addresses in order to make everyone feel that they had fast speed some of the time and then they would be able to be spun with the "shared" excuse. This is now being corrected with the implementation of a contention ratio solution. The 3 weeks is in case they stuff up and perhaps some testing time.

On a good note, I was told I will be getting the package speeds I am paying for. Well, I sure as hell hope so, but truth be told - we have heard it all before. Actions speak louder than words and right now my carrier pigeons arent very active.