Telkom: Overloading Exchanges - Proof Inside!

saffakanera

Expert Member
Joined
Sep 25, 2006
Messages
4,370
Reaction score
145


Centurion: Eco Park.

Tried Afrihost and Web Africa, getting 1000ms pings to LOCAL services. Above link was the lowest latency I got.

Called Web Africa, they advised that the exchange is overloaded.

Sigh, can someone tell me what the hell is going on with Telkom again?
 
Called Web Africa, they advised that the exchange is overloaded.

Sigh, can someone tell me what the hell is going on with Telkom again?

My question is...How does Web Africa know this. It is not information that is freely available to the public and can only be divulged by someone inside Telkom to Web Africa.

I have found over the last few weeks that Afrihost in particular has much higher latency than for example Axxess on the same ADSL service. Does this not ring bells that the ISP's actually has problems of their own and is just looking for the easiest way out by blaming the main service provider they piggy back on...Just a thought...

Yes there is a number of exchanges that is most propably over utilised but in Eco Park I do not think so because it is one of the newly developed areas in Centurion with a lot of businesses and corporate companies. So unless there is a specific system fault affecting the clients I personally can not see how it can be an overutilisation.

Just my thoughts on this...
 
I have a 384kbs line / Afrihost ISP unshaped account and my line keeps dropping spontaneously in one of the two following ways:

(less often) either unsyncing completely (which i can see by the light on my modem) or
(more often) PPPoE on red (internet) interface on my IPCOP firewall goes down, redials Afrihost, and goes back up again.

The line for some reason seems to unsync completely more often (if not only) on weekends while my red interface goes down quite spontaneously.

Ive tried:
1) Using a different modems (set up in bridge mode as my firewall is dialling)
2) Asking Telkom to reset my port (more than twice)
3) Changing all the LAN cards in my firewall for new ones

The local Telkom tech has been out three times and has tried the following:
1st time he relocated my port at the exchange. (sorry if terminology is incorrect)
2nd time he checked all the wiring outside and around my flat rewiring (he said) all the potentially faulty connections.
3rd time he again changed my location at the exchange (again not sure exactly how you word this) but also this time suggested I use a modem (like his) that has both a dsl (sync) as well as an internet light because (he said) if the internet light goes out I know that its a problem with Telkoms servers and not the line? (I did not understand this and also don’t have a modem with both lights but have as I said tried a different modem to no avail)

Has anyone had a similar problem and or maybe have any suggestions?

The line and or red interface do not go down so often that the internet is unusable but it is annoying and I don’t see why it should be going down at all…

Thanks.
 
Last edited:
1st time he relocated my port at the exchange. (sorry if terminology is incorrect)
3rd time he again changed my location at the exchange (again not sure exactly how you word this)
What they basically do is move you to another port on the same card at the exchange or a different card on the DSLAM if the one your service is working on is faulty.

2nd time he checked all the wiring outside and around my flat rewiring (he said) all the potentially faulty connections.

This can mean he found some dry joints or connections in the DP or SDC that seemed to be suspcious and not making a good connection

but also this time suggested I use a modem (like his) that has both a dsl (sync) as well as an internet light because (he said) if the internet light goes out I know that its a problem with Telkoms servers and not the line? (I did not understand this and also don’t have a modem with both lights but have as I said tried a different modem to no avail)

The modem he was referring to is one that shows an Internet connection as well as a DSL sync light. Where this basically helps is to see if your intternet connection or line sync drops. the Netgear routers sold by Telkom currently has this feature but then again if you set your router up in brdige mode the internet light means nothing because you will see your connection drop on your desktop screen when a dialup window suddenly appears.

Seeing that your connection more than often drops over weekends points to the ammount of traffic inrease on the net over weekends when all the school kids and their parents surf the net and plays online games that will cause a bottlenek at the exchange on the DSLAM if there is not enough bandwidth available on the DSLAM to make sure that all the users have sufficient resources available to use their ADSL service.

What ticks me off is that this is a well known fact and nothing seems to get done about it.
 
Thanks.

I also phoned Afrihost last week who checked my line (somehow) and said they could see (where apparantly Telkom could not?) that there was noise on the line; because they said it was running at 381 ans not 384kbps.

it does make sense what you say it as it seems to always start palying up on Sundays.

so if i inderstand you right, my firewall, PPPoE on red intertface, is basically acting as would otherwize be the internet light if i was not in bridge mode... (as an indicator when it goes down at least)

if it was just the internet light going off while the dsl snc stayed on; any ideas what that might indicate?

is there anyting I can do (that I havent already) to get it resolved?
 
Last edited:
My question is...How does Web Africa know this.
The ISPs have access to some sort of interface that Telkom provides. Nobody talks about it (Probably NDA'd) & its not clear what info they can see, but exchange info sounds plausible.

Alternatively the person he spoke to just picked it as a plausible story. We can't really know either way.
 
Here I am expecting a picture of the exchange with not even 1 port open.

a ping test is hardly any proof.
 
Here I am expecting a picture of the exchange with not even 1 port open.

a ping test is hardly any proof.

cad1aafb8d2af5.jpg
 
Now that's neat.
If they used a slightly more expensive, deeper cabinet then they could have enclosed the cabling and reduced the risk of a cable being 'accidentally' pulled out.

If you think the overhead cable trays are neat then you should lift the floor tiles and see the cables underneath the false floor!!!
 
whats your SNR and line attenuation readings on your router?

Modem is set up in bridge mode with my firewall doing the dialing. Here is a section from the log report thouth:

12:51:19 pppd[14646] No response to 3 echo-requests
11:36:04 pppd[32036] Connection terminated.
11:36:01 pppd[32036] Script /etc/ppp/ip-down finished (pid 14144), status = 0x0
11:36:01 pppd[32036] sent [LCP TermReq id=0x3 "Peer not responding"]
11:35:58 pppd[32036] sent [LCP TermReq id=0x2 "Peer not responding"]
11:35:58 pppd[32036] Couldn't increase MRU to 1500
11:35:58 pppd[32036] Couldn't increase MTU to 1500
11:35:58 pppd[32036] Script /etc/ppp/ip-down started (pid 14144)
11:35:58 pppd[32036] Serial link appears to be disconnected.
11:35:58 pppd[32036] No response to 3 echo-requests

Ive left out a large sections as prob too much to post here, but tthe above is the loop that keeps repeating every time in goes down...
 
Modem is set up in bridge mode with my firewall doing the dialing. Here is a section from the log report thouth:

12:51:19 pppd[14646] No response to 3 echo-requests
11:36:04 pppd[32036] Connection terminated.
11:36:01 pppd[32036] Script /etc/ppp/ip-down finished (pid 14144), status = 0x0
11:36:01 pppd[32036] sent [LCP TermReq id=0x3 "Peer not responding"]
11:35:58 pppd[32036] sent [LCP TermReq id=0x2 "Peer not responding"]
11:35:58 pppd[32036] Couldn't increase MRU to 1500
11:35:58 pppd[32036] Couldn't increase MTU to 1500
11:35:58 pppd[32036] Script /etc/ppp/ip-down started (pid 14144)
11:35:58 pppd[32036] Serial link appears to be disconnected.
11:35:58 pppd[32036] No response to 3 echo-requests

Ive left out a large sections as prob too much to post here, but tthe above is the loop that keeps repeating every time in goes down...

Post your line stats
 
SNR Margin Downstream= 39.0dB Upstream = 31.5dB

Line Attention = 17.0dB Up = 9.0dB

CRC Errors = 0 0

Latency = Fast Fast

ADSL parameters
Upstream 123700
Downstream 383800
 
Last edited:
SNR Margin Downstream= 39.0dB Upstream = 31.5dB

Line Attention = 17.0dB Up = 9.0dB

CRC Errors = 0 0

Latency = Fast Fast

ADSL parameters
Upstream 123700
Downstream 383800

That is some insanely good stats right there...

The fact that you sync at 381k instead of 384k has nothing to do with your line, ADSL2+ does that sometimes. If you force your modem to G.DMT it will sync at 384k. With those stats however you should have the best 384k experience ever. Chances are good your exchange is indeed running at max capacity.

Post a traceroute.

If your exchange is not the problem, only other problem is, something on your pc or network is running the line at max speed 24/7 and that is why your latency is so high. On 384k its too easy to run a line at max speed, not to mention some malware could be doing this for you.

A traceroute should help us identify if the problem is something hogging your line or if your exchange is overloaded.
 
C:\Users\jlct021>tracert www.google.com

Tracing route to www.l.google.com [196.23.168.147]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms ipcop.localdomain [172.18.91.150]
2 26 ms 30 ms 29 ms 196-210-148-129.dynamic.isadsl.co.za [196.210.14
8.129]
3 51 ms 51 ms 47 ms cdsl1-rba-vl2460.ip.isnet.net [196.38.72.217]
4 53 ms 48 ms 51 ms cdsl1-rba-vl150.ip.isnet.net [196.38.73.17]
5 47 ms 51 ms 51 ms 168.209.1.139
6 48 ms 49 ms 51 ms 168.209.1.188
7 46 ms 51 ms 50 ms za196-23-168-147.cache.googlevideo.com [196.23.1
68.147]

Trace complete.

C:\Users\jlct021>
 
Last edited:
Top
Sign up to the MyBroadband newsletter
X