Crystal Web (NEW ISP) ADSL Feedback

Status
Not open for further replies.
For those of you asking "why is there packet loss at my exchange on your account but not on <insert other ISP>"?

The bottom line? We have no cooking clue. It could be one of a million reasons - none of which have anything to so with us, and all of which have to do with Telkom. i.e. Telkom's config is wrong on the port, or there is a very specific hardware fault, or etc.

You have to remember, we are not Telkom, and we have zero control over their infrastructure. We also cannot directly test the equipment at their exchanges. If you have congestion at the exchange, it will not help asking our agents to explain the technical nature of the congestion, nor asking them for the technical explanation as to why on Telkom's network you are seeing different results under differing circumstances. You are asking the wrong company - we simply do not know; we cannot help fix it; and we cannot diagnose the problems either. We'd love to, but we can't, I'm afraid.

If we have spent the time supporting, and have identified the fault, and request that it be reported to Telkom, your best course of action is to report it to Telkom with the data and info we supply you with. We'll help all we can there. But we cannot help to diagnose the root-cause of the issue, nor assist to fix the problem. It's like asking South African Airways why the food on Kulula flights is often dry. SAA can speculate to their heart's content, but it will serve absolutely no purpose. The question you're posing should be directed to Kulula's support channels instead.
 
To be fair, Deezer on other ISPs (from MWEB servers) just barely stays ahead of the play bar. It's not exactly coming down full speed, but it is fast enough.

And prioritization for QoS is fine, but when low priority goes with packet loss it becomes rather bad.

This traceroute was running just now over the last few minutes. They were started at the same time. Just adding more info to the pile and really hoping there's some switch that someone forgot to flick or something.

Capture.PNG
 
To be fair, Deezer on other ISPs (from MWEB servers) just barely stays ahead of the play bar. It's not exactly coming down full speed, but it is fast enough.

And prioritization for QoS is fine, but when low priority goes with packet loss it becomes rather bad.

This traceroute was running just now over the last few minutes. They were started at the same time. Just adding more info to the pile and really hoping there's some switch that someone forgot to flick or something.

View attachment 156126

This is where it's good to get some decent data. 105.233.62.1 is your exchange. Average and best look good - worst is 137.4ms though. Resulted in packet loss immediately afterwards. It's only a single data point so I'm not going to point too many fingers there, but the second test confirms your exchange is fluctuating considerably right now in terms of performance. And that's only running over 400 odd packets, so it's happening consistently in that case.

There may be a case to be had for us to look at ways to optimise the QoS to be less susceptible to incoming packet loss issues, but it's built into the packets sent to us and manipulating anything there would be downright illegal, and optimising the pipe would mean optimising the pipe-size essentially - i.e. increasing costs. Should we be carrying that burden for poor exchanges?

Also, some of the tests being posted here are flouting the lines a little and should best be kept to tests for the sake of curiosity rather than network performance.

We do encourage real world testing rather than speedtests in a similar vein.
 
but the second test confirms your exchange is fluctuating considerably right now in terms of performance.

I'd like to understand this better. I don't know anything about what happens at the exchange except that I'm far away from it (50db attenuation) so getting 4mbps would be a miracle. Across the road is 40mbps, but hey.

The two traceroutes were running at the same time, on different ISPs but over the same copper. What happens at the exchange to let it "fluctuate" on one path only? If I didn't know better I would say someone needs to go clean one of the network ports, but I now get the idea that each ISP has a specific pipe size at each exchange and this might mean that a lot of people in my area are using you (or Cybersmart) as ISP? Every time I move into a new home I redo ISP comparisons and each time I found a different ISP to be better.
 
I'd like to understand this better. I don't know anything about what happens at the exchange except that I'm far away from it (50db attenuation) so getting 4mbps would be a miracle. Across the road is 40mbps, but hey.

The two traceroutes were running at the same time, on different ISPs but over the same copper. What happens at the exchange to let it "fluctuate" on one path only? If I didn't know better I would say someone needs to go clean one of the network ports, but I now get the idea that each ISP has a specific pipe size at each exchange and this might mean that a lot of people in my area are using you (or Cybersmart) as ISP? Every time I move into a new home I redo ISP comparisons and each time I found a different ISP to be better.

Both results on the two ISP accounts show high latency at the exchange - more so on the other ISP, in fact. Capacity isn't purchased at the exchanges, by the way - it's purchased at central IPC points further down the line than the exchange, which is still on Telkom's last mile infrastructure - in other words it is part of what we as an ISP pay for access to. So in essence, neither you, nor your ISP are getting what we're paying for when the exchange is congested, and furthermore we take the brunt of our support agents spending hours on end every day diagnosing this nonsense.
 
Capacity isn't purchased at the exchanges, by the way - it's purchased at central IPC points further down the line than the exchange, which is still on Telkom's last mile infrastructure - in other words it is part of what we as an ISP pay for access to. So in essence, neither you, nor your ISP are getting what we're paying for when the exchange is congested, and furthermore we take the brunt of our support agents spending hours on end every day diagnosing this nonsense.
Build some POPs? :p
 
Build some POPs? :p

As tongue-in-cheek as that is for a small ISP like us to do, it's not a bad suggestion. Now for us to secure national traffic and to invest in the last mile before....wait, what's that? Fibre?

:p

Or, we should demand that we get what we pay for. There's still a case for tiered players in this market.
 
Last edited:
Or, we should demand that we get what we pay for. There's a still a case for tiered players in this market.
I agree and I think that with TelkomInternet moving away from SAIX and also going the IPC route, it will benefit all ISPs as TI will now be putting massive pressure on Telkom Wholesale to up their game.
 
I agree and I think that with TelkomInternet moving away from SAIX and also going the IPC route, it will benefit all ISPs as TI will now be putting massive pressure on Telkom Wholesale to up their game.

With the growing pressure of alternatives to IPC and the lowered cost of provisioning non-last mile connectivity over alternative technology, Telkom really should be looking to secure their wholesale business better right now, rather than their hellbent intent on direct consumer relationship ownership - the latter being a model that's seen the dirtiest, most unsustainable, and downright silliest market moves by them over the years. TI is certainly a good step forward in that respect but those Chinese walls are hardly the barrier they should be right now.

This market is going to see great growth over the years (not necessarily all provisioned over DSL) but the unfortunate bit with respect to DSL is that Telkom Wholesale don't appear to be leading the charge, which in turn will secure their continued market position in a competitive environment.
 
So in essence, neither you, nor your ISP are getting what we're paying for when the exchange is congested
So in essence, neither you, nor your ISP are getting what we're paying for even when the exchange is not congested.

Fixed that for you.
 
Is this an accurate summary?

I can run traceroute for 3 different ISP's, in cluding CW.

I can do the tests 10 times and everytime it is only CW that shows packet loss.

CW does not know how rectify this.

It can possibly be rectified by spending more money, but it is not going to happen.
 
Is this an accurate summary?

I can run traceroute for 3 different ISP's, in cluding CW.

I can do the tests 10 times and everytime it is only CW that shows packet loss.

CW does not know how rectify this.

It can possibly be rectified by spending more money, but it is not going to happen.

Wow. If that's the impression that you've got then I need to quit doing this now.

Exchange congestion is not something we can fix. It is not our network. One of the solutions to helping Telkom themselves with their plethora of exchange congestion issues would be for ISPs to purchase massive IPC and international bandwidth instead so as to reduce the issues created by packet loss at an exchange. It won't resolve it - it will only act as a temporary band aid for some customers, and will still massively affect the vast majority on that exchange irrespective of how much capacity we have.

So no, I'd say that your summation is wholeheartedly incorrect and quite disheartening to read to be honest.

What I meant with that previous comment that seems like it's been completely twisted, is that we would have to essentially purchase capacity to provision unshaped and minimal contention accounts to alleviate some of the pain of congested exchanges. And then sell it at current pricing. That isn't going to happen, I'm afraid.
 
Last edited:
Doesn't matter how much IPC and transit you have, if there's a bottleneck between your network and the end-user, they aren't gonna see ANY benefit to increasing it.
 
CW does not know how rectify this.

This part is also quite difficult to swallow to be honest. What would customers propose we do as an ISP to solve your issues with your Telkom last mile infrastructure? We cannot test it. We cannot change it. We cannot do anything with it - nothing. It's not ours.

I feel like there is a lot of anger from people across all ISPs about this sort of thing, and I feel your pain. The anger is often directed at the ISP though, which is not right. We're neither the cause, nor the solution for the problem in cases of congested exchanges, and moving forward, we are now going to have to rethink quite a few things with respect to how we interact on this basis.
 
Doesn't matter how much IPC and transit you have, if there's a bottleneck between your network and the end-user, they aren't gonna see ANY benefit to increasing it.

I'd imagine it may help a few % of customers on the fringes of congestion. But it would be a drop in the ocean overall.
 
This part is also quite difficult to swallow to be honest. What would customers propose we do as an ISP to solve your issues with your Telkom last mile infrastructure? We cannot test it. We cannot change it. We cannot do anything with it - nothing. It's not ours.

I feel like there is a lot of anger from people across all ISPs about this sort of thing, and I feel your pain. The anger is often directed at the ISP though, which is not right. We're neither the cause, nor the solution for the problem in cases of congested exchanges, and moving forward, we are now going to have to rethink quite a few things with respect to how we interact on this basis.
Okay.
I'm not attacking , I was asking.

My previous post was factually correct for me.

I can test 3 ISP's and CW would be the only ISP's with packet loss.

The other points were made, by you, but Gary's list probably comes in to play ito I quoted you out of context.

You were talking mostly about exchanges and the difficulties around it.
I was never thinking about exchanges, only about my own situation.

I apologize for using your words in that context. It was not fair or correct.

Fact is though, I'm sitting with packet loss and there have been several promises and none of then have panned out.

I live in hope. I like my account and hope the problems can be sorted for me in the EC.
 
You would still need to install some equipment at each site else it will still not address the congestion issue.
yep but the point is that getting facilities in the form of buildings, access to power etc ... are really important here. There should be costs involved - reasonable lease charges.
 
Status
Not open for further replies.
Top
Sign up to the MyBroadband newsletter
X