Afrihost Uncapped ADSL Feedback

Status
Not open for further replies.
Had the crappiest speed since last night already. Will try and reboot when I get home. Hope that will help. I guess that "uncapped promo" couldn't have come at the worst time...
 
Telkom Cappend Unshaped:

Test conducted on 14 March 2013 10:47:57

Local:
Download Speed: 7594 kbps (949.3 KB/sec transfer rate)
Upload Speed: 843 kbps (105.4 KB/sec transfer rate)
Latency: 41 ms

FNB free account:

Test conducted on 3/14/2013 10:54:09 AM

Download Speed: 7121 kbps (890.1 KB/sec transfer rate)
Upload Speed: 848 kbps (106 KB/sec transfer rate)
Latency: 42 ms

Afrihost 10mb uncapped:

Test conducted on 3/14/2013 10:56:15 AM

Download Speed: 7692 kbps (961.5 KB/sec transfer rate)
Upload Speed: 853 kbps (106.6 KB/sec transfer rate)
Latency: 38 ms
 
kripstoe said:
Can you do a traceroute to 8.8.8.8 ?

Would like to see if you get similar problems along the way as I did.
Click to expand...

Tracing route to google-public-dns-a.google.com [8.8.8.8]
over a maximum of 30 hops:

1 4 ms 4 ms 4 ms 105-236-8-193-esr-lo.mtnbusiness.co.za [105.236.8.193]
2 28 ms 31 ms 29 ms ipc-recieve-tb-1a.mtnbusiness.net [41.181.53.150]
3 6 ms 5 ms 5 ms tb-dca-2.za--qux-a.za.mtnbusiness.net [41.181.184.28]
4 6 ms 6 ms 6 ms unc-cpt-1.mtnns.net [196.44.18.8]
5 7 ms 6 ms 6 ms 196.44.31.106
6 31 ms 32 ms 30 ms ct-cr-2.za--rb-cr-2.za-a.mtnns.net [196.44.31.30]
7 39 ms 42 ms 38 ms 41.181.139.99
8 59 ms 31 ms 31 ms 72.14.194.74
9 180 ms 180 ms 179 ms 72.14.232.102
10 221 ms 207 ms 208 ms 209.85.253.10
11 187 ms 187 ms 186 ms 72.14.232.76
12 186 ms 186 ms 186 ms 209.85.254.114
13 * * * Request timed out.
14 186 ms 186 ms 187 ms google-public-dns-a.google.com [8.8.8.8]

Trace complete.
 
We seem to have found a link on the network that was misbehaving, not really sure of the technical details. We're seeing that clients rebooting are getting good throughput now. We're not sure if this is limited to CTN users, but so far we're managing to resolve support queries on email and phones with router reboots.

Is there anyone getting no throughput at all, or not able to connect or authenticate?
 
I wonder why a SMS was sent by Afrihost about some or other problem .......... I had no problems! My nntp, p2p, email and http worked just fine!
 
AfriMan said:
This is possibly a question best directed at more senior people within MTN and our Dev department, but I can tell you that it's a very complex setup and the guys who run it are some of the smartest people around. As far as we know they use international standards and best practices, and the equipment used is the very top of the range.
Click to expand...

Well, it makes me wonder since you had this "back-log" of authentication requests once before. Did these smart people not then look at possible future scenarios to avoid it from happening.

I do understand that it may seem overly simple to some, but I am confident that this is not due to oversight or negligence on their part,
Click to expand...

So are you now in a position to explain to us, or relay from those more senior people what happened yesterday. Purely speculation here, but it sounds like Afrihost's radius environment struggled to handle all the incoming requests when your IPC came back online, and that caused the MTN proxies to take a severe beating.

It would be nice for once if you can tell us exactly what happened, and what measures you now have in place to avoid being down for most of a day going forward.

Surely it's not too much to ask?


Michael
 
mikeygza said:
Well, it makes me wonder since you had this "back-log" of authentication requests once before. Did these smart people not then look at possible future scenarios to avoid it from happening.



So are you now in a position to explain to us, or relay from those more senior people what happened yesterday. Purely speculation here, but it sounds like Afrihost's radius environment struggled to handle all the incoming requests when your IPC came back online, and that caused the MTN proxies to take a severe beating.

It would be nice for once if you can tell us exactly what happened, and what measures you now have in place to avoid being down for most of a day going forward.

Surely it's not too much to ask?


Michael
Click to expand...

+1

Would like to know myself...
 
pgs said:
I wonder why a SMS was sent by Afrihost about some or other problem .......... I had no problems! My nntp, p2p, email and http worked just fine!
Click to expand...

I think the SMS went out to all ADSL clients, as we're still not 100% certain which clients were affected! Glad you weren't interrupted though :)
 
AfriMan said:
I think the SMS went out to all ADSL clients, as we're still not 100% certain which clients were affected! Glad you weren't interrupted though :)
Click to expand...

Would like to confirm that I received the SMS - had to assume that it is because I've still got that 1Gb free account :)
 
AfriMan said:
I think the SMS went out to all ADSL clients, as we're still not 100% certain which clients were affected! Glad you weren't interrupted though :)
Click to expand...

and sending an sms after the fact to apologise helps how exactly?

I think you guys need a course in public relations management
 
I'm currently on a 1Mb line that is used 99% of the time in the evenings. I'm considering upgrading to a 4Mb line but these constant issues concern me.
 
mikeygza said:
Well, it makes me wonder since you had this "back-log" of authentication requests once before. Did these smart people not then look at possible future scenarios to avoid it from happening.



So are you now in a position to explain to us, or relay from those more senior people what happened yesterday. Purely speculation here, but it sounds like Afrihost's radius environment struggled to handle all the incoming requests when your IPC came back online, and that caused the MTN proxies to take a severe beating.

It would be nice for once if you can tell us exactly what happened, and what measures you now have in place to avoid being down for most of a day going forward.

Surely it's not too much to ask?


Michael
Click to expand...

At this stage I don't have all the information. From what I do know, there was a breakdown in the authentication chain between Telkom - MTN - Afrihost, and authentications were not being correctly parsed. We had to try a few things to get users who were backlogged authenticated, and some of them were being rejected without reaching our radius. Nothing fell over as far as I know, but somewhere there was a bottle neck, and MTN are still working with Telkom to find out what happened. They flushed the queue several times, and also initiated some workarounds, which eventually got most people on, but then found that the issue was somewhere between Telkom and MTN and the information parsed there. This was then fixed on one of our network nodes and tested, and once we saw that authentications started going through smoothly, we replicated the fix across the network.

We then mass disconnected users to force them re-authenticate so that those whose authentication was affected (who had an IP and no throughput) would authenticate correctly and be able to to get normal throughput. Some users were not able to be disconnected remotely, and a small percentage could only be reset physically on their routers.

As far as we know, it was not systems falling over due to the amount of requests, but inconsistencies in the parsed information that wouldn't allow the queue to resolve as it normally would have. Our systems, and MTN's, are designed to anticipate the unlikely event of every user on the network authenticating at the same time or over a short period. So it was not due to poor planning or inferior infrastructure. We're still investigating with MTN, but right now the priority is integrity testing and performance monitoring to make sure everything is as it should be.
 
Ol' Mean Bastid said:
and sending an sms after the fact to apologise helps how exactly?

I think you guys need a course in public relations management
Click to expand...

The SMS is an apology, we updated clients as best we could during the day over social media and here on MyBroadBand. We're of the mind that every bit helps and we understand that some people still feel angry about the issue :(
 
Just another thank you to Afrihost quick response and assistance last night.

Also got the apology SMS this morning :D

Going to test it tonight, hopefully I can get the congestion Telkom issue sorted out as well.
 
So has Afrihost stablized in durban since the MTN move or is it still running poorly???

Thinking of changing from mweb to afrihost.

Mweb 2mb uncapped 599 incl line rental. (afrihost 497pm incl line rental)
Afrihost 4mb 697 inclu line rental (mweb 4mb @ 899pm incl line rental...

Really wondering if mwebs that much better
 
Last edited:
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X