Afrihost - Pure Fibre Feedback Thread Part 2

jimini said:
@Afrigirha @AfriNatic

I have now lost connection completely - Vuma Villages / Lonehill.

This is becoming a joke.
Click to expand...
Hi.

Network Incident has been logged and the repairs are in progress.
 

Attachments

  • Screenshot_20230123-073253_Chrome.jpg
    Screenshot_20230123-073253_Chrome.jpg
    144 KB · Views: 23
East London Clients.
@username and 23 others @K T @xbxbau5 @waynehooper
I have submitted an escalation to Vuma's Service Manager and their NOC Team with the latest results. There has not been a Network Incident logged because Vuma does not believe there is a problem in the area; they have requested test results, and they will go through them and provide feedback. If anyone else in EL is experiencing degraded services during peak time, please share your results, and I will submit them to Vuma. This should push them to investigate further.
 
Afrigirl said:
East London Clients.
@username and 23 others @K T @xbxbau5 @waynehooper
I have submitted an escalation to Vuma's Service Manager and their NOC Team with the latest results. There has not been a Network Incident logged because Vuma does not believe there is a problem in the area; they have requested test results, and they will go through them and provide feedback. If anyone else in EL is experiencing degraded services during peak time, please share your results, and I will submit them to Vuma. This should push them to investigate further.
Click to expand...
Show them this - it shows the problem started on 30 December as well as the nightly service issues. I will email you the latest raw data from the Asus-Merlin Connmon app showing the times when issues occur.

Screenshot 2023-01-23 113554.jpg
 
Soul Assassin said:
Thank you, if it was just a visual thing I wouldn't have bothered mentioning it but the ping in game also goes up the additional 20ms when the path changes and I don't remember it always being like this.

It was a solid ~170ms for the longest time and now it's on the 190ms for a couple of days and then 170ms just for a couple of hours before going back to 190ms.
Click to expand...

Hi

We had a 1-hour meeting today to discuss the issue and possible solutions but unfortunately there isn't an immediate solution that can be engineered right now. Let me dicuss a few things here which is causing the issue and why a solution is not possible right now.

Afrihost uses DHCP for most FNO's excluding Openserve and Vodacom. With DHCP there are static vlan configs on the cross connects from the FNO's which we specify /23 or /22 ipv4 subnets.

We use 3 main transit providers. Openserve, Cogent and Liquid Telecoms. In JHB egress is mostly Cogent while Liquid and Openserve egress has less weight applied so they will likely just be used in the case that Cogent goes down completely. KZN and CPT we prefer Openserve egress.

How networks in the EU see us depends on where they have a presence and their configs. We can unfortunately not control this. To control this we need to withdraw prefixes from our transit providers and that is not feasible due to the amount of traffic we move.

We had to add liquid telecoms to the mix because we needed more capacity and they have given us a good rate and the capacity we need. Had we gone for more expensive options a price increase would have been likely and probably given us the same result.

Now considering our traffic levels and the options available to us from transit providers it's not possible to engineer a quick fix now, but we are looking at a so-called express route from JHB to EU via London.

To achieve this we need pppoe and a /22 divided into 2 and then announce this prefix only on a specific transit provider that offers the best route which is Openserve for now. We will set it for forward path (egress) and the reverse (ingress) to ensure it stays consistent. For this, to work we need to apply a static IP to a client from these subnets so it needs some dev work to engineer and can work once we have our migration to pppoe completed. Openserve and Vodacom FNO's will require extra work to get them to work and it will likely only be a nat static type setup.
 
Afrigirl said:
East London Clients.
@username and 23 others @K T @xbxbau5 @waynehooper
I have submitted an escalation to Vuma's Service Manager and their NOC Team with the latest results. There has not been a Network Incident logged because Vuma does not believe there is a problem in the area; they have requested test results, and they will go through them and provide feedback. If anyone else in EL is experiencing degraded services during peak time, please share your results, and I will submit them to Vuma. This should push them to investigate further.
Click to expand...
Here's a whole thread filled with graphs over the last few weeks.

Vumatel East London

Still no change on my end. Any thoughts on whether frogfoot is any better I would also like to know about Frogfoot. My line died this afternoon, now waiting for Vuma to repair
mybroadband.co.za mybroadband.co.za

If they still don't believe it's a problem I can be belligerent
CONsKvh8mx.png
 
Noah said:
Here's a whole thread filled with graphs over the last few weeks.

Vumatel East London

Still no change on my end. Any thoughts on whether frogfoot is any better I would also like to know about Frogfoot. My line died this afternoon, now waiting for Vuma to repair
mybroadband.co.za mybroadband.co.za

If they still don't believe it's a problem I can be belligerent
View attachment 1462409
Click to expand...
Hi @Noah, if Afrihost manages your line, please send me a PM with your email address, so I can get your line number to link to your test results.
 
AfriNatic said:
Hi

We had a 1-hour meeting today to discuss the issue and possible solutions but unfortunately there isn't an immediate solution that can be engineered right now. Let me dicuss a few things here which is causing the issue and why a solution is not possible right now.

Afrihost uses DHCP for most FNO's excluding Openserve and Vodacom. With DHCP there are static vlan configs on the cross connects from the FNO's which we specify /23 or /22 ipv4 subnets.

We use 3 main transit providers. Openserve, Cogent and Liquid Telecoms. In JHB egress is mostly Cogent while Liquid and Openserve egress has less weight applied so they will likely just be used in the case that Cogent goes down completely. KZN and CPT we prefer Openserve egress.

How networks in the EU see us depends on where they have a presence and their configs. We can unfortunately not control this. To control this we need to withdraw prefixes from our transit providers and that is not feasible due to the amount of traffic we move.

We had to add liquid telecoms to the mix because we needed more capacity and they have given us a good rate and the capacity we need. Had we gone for more expensive options a price increase would have been likely and probably given us the same result.

Now considering our traffic levels and the options available to us from transit providers it's not possible to engineer a quick fix now, but we are looking at a so-called express route from JHB to EU via London.

To achieve this we need pppoe and a /22 divided into 2 and then announce this prefix only on a specific transit provider that offers the best route which is Openserve for now. We will set it for forward path (egress) and the reverse (ingress) to ensure it stays consistent. For this, to work we need to apply a static IP to a client from these subnets so it needs some dev work to engineer and can work once we have our migration to pppoe completed. Openserve and Vodacom FNO's will require extra work to get them to work and it will likely only be a nat static type setup.
Click to expand...
I always appreciate the transparency and a peak behind the curtain, I'm back on the 170ms as of an hour ago, hopefully it sticks around a bit longer this time.

One question if you're allowed to say, when was Liquid added into the mix?
 
Soul Assassin said:
I always appreciate the transparency and a peak behind the curtain, I'm back on the 170ms as of an hour ago, hopefully it sticks around a bit longer this time.

One question if you're allowed to say, when was Liquid added into the mix?
Click to expand...

It's just over 2 months or so. We had to implement it in stages starting with CPT and KZN and lastly in JHB.
 
AfriNatic said:
Hi

We had a 1-hour meeting today to discuss the issue and possible solutions but unfortunately there isn't an immediate solution that can be engineered right now. Let me dicuss a few things here which is causing the issue and why a solution is not possible right now.

Afrihost uses DHCP for most FNO's excluding Openserve and Vodacom. With DHCP there are static vlan configs on the cross connects from the FNO's which we specify /23 or /22 ipv4 subnets.

We use 3 main transit providers. Openserve, Cogent and Liquid Telecoms. In JHB egress is mostly Cogent while Liquid and Openserve egress has less weight applied so they will likely just be used in the case that Cogent goes down completely. KZN and CPT we prefer Openserve egress.

How networks in the EU see us depends on where they have a presence and their configs. We can unfortunately not control this. To control this we need to withdraw prefixes from our transit providers and that is not feasible due to the amount of traffic we move.

We had to add liquid telecoms to the mix because we needed more capacity and they have given us a good rate and the capacity we need. Had we gone for more expensive options a price increase would have been likely and probably given us the same result.

Now considering our traffic levels and the options available to us from transit providers it's not possible to engineer a quick fix now, but we are looking at a so-called express route from JHB to EU via London.

To achieve this we need pppoe and a /22 divided into 2 and then announce this prefix only on a specific transit provider that offers the best route which is Openserve for now. We will set it for forward path (egress) and the reverse (ingress) to ensure it stays consistent. For this, to work we need to apply a static IP to a client from these subnets so it needs some dev work to engineer and can work once we have our migration to pppoe completed. Openserve and Vodacom FNO's will require extra work to get them to work and it will likely only be a nat static type setup.
Click to expand...
No idea what you just said, but this kind of feedback should be an example to other providers.
Thank you.
 
Another night of latency
Screenshot 2023-01-23 200009.jpg
14249443365.png
 
K T said:
Here we go again! Same here my side. Another night wasting at least half an hour doing power-offs trying to troubleshoot/resolve the issue so I can get some service.

Or maybe I should just wait until 10PM
Click to expand...
Oh, and by the way, it's not chump change that I'm paying for the service that I don't have, at the time when I want to just sit down and use it.
 
K T said:
Oh, and by the way, it's not chump change that I'm paying for the service that I don't have, at the time when I want to just sit down and use it.
Click to expand...
That’s the problem, they buy out the competition and can give us kak service. We just have to bend over and take it. I have downgraded my speeds so they will be seeing less of my money. Any more of this and they will see none
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X