LATENCY!!!

what planned telkom maintenance???

@ EVERYONE please will you phone telkom and harass them about the problem (do so again, if you already have) until they do something about it.

phone:
10217 option 2, then 3
and 08600 375375
and 10212

get put through to a technion or manager if possible.
tell them there is are latency issues and packet loss in all the areas where 10 meg upgrades occurred 2 months ago. its an ongoing and widespread problem which started at the beginning of october and is affecting 100s of people. 1) are they aware of the problem? 2) what is the cause? 3) how are they planning to fix it and when will it be fixed? 4) if all else fails suggest that the 10 meg upgrades caused the problems and demand they downgrade everyone back to 4 meg (at least temporarily to see if it helps).


@ Will and Jeff... i assume you guys have contacts higher up at telkom since u are big ISPs and have to work with those evil bastards a lot. if not, then im sure theres someone at your ISP who deals with telkom directly. tell them to get their **** together and sort this ****ing **** out NOW!!!

LOL I've already been through 3 fault logs - the first they moved me to a different DSLAM, the second they switched the line from the exchange to my house, the third they moved me back to my old DSLAM. In each case - it was TICK job done, fault closed. I'll be opening another tomorrow.
 
they upgraded everyone to 10 meg in a really short period of time

Dude. They ran a trial for faster than 4 meg (i.e. 10 meg) for a while...
 
Is there not some option to create something that ping sensitive customers can be switched to at the cost of the "10 meg capable" piece of $#%^ I've got right now?

Anyone know if ISDN is affected, because that seriously an option for me right now...

1. That's one of the reasons that telcos in the US would like to be able to prioritise customers based on price, but are limited by the 'common carrier' and 'net neutrality' rules (google if you want more).

2. Conspiracy theorists would say that is exactly Telkom's strategy, to reduce ADSL to the lowest common denominator, and charge a premium service for latency sensitive customers. These are mostly business and gamers (guess who's willing to pay more...)

3. peTerp00n: I can understand your frustration, but 10Mbps is not really the issue, nor a viable solution IMO.

4. Does anyone know how many people are running DSTV via IP, since this could be hogging a lot of bandwidth?
 
are you just saying that because mweb now have that really expensive 10 meg uncapped product? i really dont understand, they upgraded everyone to 10 meg in a really short period of time, no questions asked. so why would it be a mammoth undertaking to do the reverse? what problems would it cause? personally i think it would solve all our problems.

No definitely not :D Just that I know that there was a lot more to the rollout than simply flicking a switch for some users to be on 10mb and in a lot of cases the work done was also upgrading the capacity of aging infrastructure. Really regression is not an option here, they need to find the cause of our problems, acknowledge the issue and fix it - it really is that simple.
 
I split my graph up as well (thanks Gordon - didn't even realise Ping Plotter could do that) - latency problem is all between hops 2 and 3 (which we know but interesting to see how little latency problem there is after that). However, packet losses are mostly towards the end, and especially the last hop I've noticed http://www.imagehost.co.za/share-1DC5_4CC3DBC0.html

It's sad, but I would even consider paying a premium to be connected to a 'priority' DSLAM. I'm also about to risk my life and climb to the peak of my roof to see if I have a clear line of sight to the building in Claremont that houses my company's IT providers. Then I can set up a wireless connection to their Neotel line, which isn't having these problems.
 
I split my graph up as well (thanks Gordon - didn't even realise Ping Plotter could do that) - latency problem is all between hops 2 and 3 (which we know but interesting to see how little latency problem there is after that). However, packet losses are mostly towards the end, and especially the last hop I've noticed http://www.imagehost.co.za/share-1DC5_4CC3DBC0.html

It's sad, but I would even consider paying a premium to be connected to a 'priority' DSLAM. I'm also about to risk my life and climb to the peak of my roof to see if I have a clear line of sight to the building in Claremont that houses my company's IT providers. Then I can set up a wireless connection to their Neotel line, which isn't having these problems.

PingPlotter has a lot of settings, although the defaults are fine for most options.

My graph is not continous, but shows very much the same trends...

Getting uniform chart size would allow us to overlay exactly, but the correlations are pretty clear IMO...

Try turning on 'graph min/max' latency, which will show you a lot more informaton (jitter).

There's a DFA fiber line running up my road, pity I can't connect to it...
 
LOL I actually had to switch off the min/max latency graphing because the jitter was so bad it squashed the rest of the graph up.

All this fibre around and adsl remains constipated.
 
I'm doing tracert to tenet.cinx.net.za, which is only 5 hops, so jitter is tolerable.

I suggested to Will that telkom.co.za was a bad choice for tracert (11 hops), but does show huge jitter. I guess both charts prove different points!

PingPlotteris quite flexible: You can select which, or both, graphs you want to select, since the dataset stores all the traces simultaneously (my file is now 500KB).

My 3rd hop IP route has changed from 196-28-178-166.adsl.mweb.co.za to 196-28-178-202.adsl.mweb.co.za, so my graph is really broken. (Also had power failure, and switched off PC overnight).

Conspiracy theorists might say that Telkom is 'laming' ADSL to push users in other directions...

Update: Latency slowly creeping up to 50ms at 10am on Sunday. Will be horrible after lunch...

Update2: Latency above 100ms by 10.30am, and fluctuating widely thereafter (mostly towards 150ms...)

P.S. I will be very happy when I can stop looking at tracert all the time, and actually use the internet 24/7!
 
Last edited:
Last edited:
I'm not sure what that was Gordon, shows up as a slight dip in the graphs on our side, but no alarms went off, I've escalated it in any event just to be on the safe side.
 
Not to be pointing any fingers, but when did M-Web go live on CINX? I found this twitter message which suggested a date, which is around the time I started having latency issues.

http://twitter.com/jacomuller/status/26635061056

Perhaps the congestions issues is to do with the increased amount of uncapped MWEB traffic since MWEB went live in CINX?
 
Not to be pointing any fingers, but when did M-Web go live on CINX? I found this twitter message which suggested a date, which is around the time I started having latency issues.

http://twitter.com/jacomuller/status/26635061056

Perhaps the congestions issues is to do with the increased amount of uncapped MWEB traffic since MWEB went live in CINX?

Exremely unlikely! The total CINX traffic (all ISPs together) is less (400Mbps) than one mediocre router could handle: http://stats.cinx.net.za/showtotal.php?img=week
There is very little demand from MWeb ADSL users to CINX, mainly TENET (UCT, etc), and a few small ISPs and hosting companies AFAIK.
 
Ah well just wasted 70 bucks on a axxess prepaid, at least now i know its definitely not mweb
 
Ah well just wasted 70 bucks on a axxess prepaid, at least now i know its definitely not mweb

Please post your latency (tracert is fine). We haven't had any from Plumstead/Wynberg for a while.

Fairly steady at 150ms in Clareinch almost the whole day...

Edit: We haven't had many SAIX users on this thread. Either they haven't noticed, or Telkom has a 'special router' for its own customers (wouldn't surprise me at all...)
 
Last edited:
I've just started my pingplotter but i'll quickly get traces for both accounts
MWEB
Code:
Tracing route to tenet.cinx.net.za [198.32.214.33]
over a maximum of 30 hops:

  1     1 ms     2 ms     2 ms  rb.toady.ctwug.za.net [172.18.42.110]
  2     8 ms    25 ms    34 ms  41-132-48-1.dsl.mweb.co.za [41.132.48.1]
  3   158 ms   139 ms   158 ms  196-28-178-166.adsl.mweb.co.za [196.28.178.166]

  4   173 ms   199 ms   189 ms  196-28-178-133.adsl.mweb.co.za [196.28.178.133]

  5   139 ms   151 ms   151 ms  tenet.cinx.net.za [198.32.214.33]

Trace complete.

Axxess PrePaid
Code:
Tracing route to tenet.cinx.net.za [198.32.214.33]
over a maximum of 30 hops:

  1     3 ms     2 ms    22 ms  rb.toady.ctwug.za.net [172.18.42.110]
  2     7 ms    15 ms     9 ms  196-210-152-1.dynamic.isadsl.co.za [196.210.152.
1]
  3   188 ms   179 ms   177 ms  196.38.73.113
  4   188 ms   194 ms   200 ms  cdsl2-rba-vl150.ip.isnet.net [196.38.73.9]
  5   170 ms   165 ms   169 ms  196.26.0.61
  6   202 ms   200 ms   195 ms  core3a-ctn-gi0-3-90.ip.isnet.net [168.209.100.15
4]
  7   204 ms   205 ms   208 ms  cp2-ctn-gi0-1.ip.isnet.net [168.209.2.14]
  8   188 ms   191 ms   195 ms  tenet.cinx.net.za [198.32.214.33]

Trace complete.
 
Last edited:
I've just started my pingplotter but i'll quickly get traces for both accounts

That's pretty conclusive in my books...

BTW: Your whole MWeb trace (2nd hop onwards) is identical to mine, and I'm on Clareinch. Not sure what that means...

P.S. IS ADSL in CT doesn't seem to peer on CINX (yet).
 

Very similar to my chart today, except that mine (Clareinch) had a dramatic dip at 13h40. Seems Rondebosch was not affected by that dip (Will suggested it was not widespread).

What's quite puzzling is that your 2nd hop (41-132-48-1.dsl.mweb.co.za) is the same mine, as well as acidrain in Plumstead.

Will: Is it possible that we are all on the same DSLAM or ESR??
 
Top
Sign up to the MyBroadband newsletter
X