South Africa’s biggest forum. Discuss, discover, and connect with thousands of members.
Technically he is right. High demand could result in an account feeling "shaped". Either way, it is absolute crap. I feel sorry for Afriguy, but why pretend everything is A-OK?
Agreed.
The Bible also says that the universe was created in 6 days..
Afriman, claiming something to be true is not proof that it's true. Surely, if afrihost are not shaping, then you should have no problem in dismissing these allegations?
Because as things stand, all the evidence seems to suggest otherwise. In fact, the only proof that you can offer us is your word..
I'm no expert so I may be completely wrong but if high demand was causing the account to feel shaped surely it would affect all protocols and not just the ones that are traditionally shaped?
The way I see it is for e.g if there is an accident on the highway that causes a bottleneck it will cause all lanes to slow down and not leave the right lane at full speed while the left lane hardly moves?
We definitely know that everything is not perfect and I don't believe we've tried to say this. We know that we're struggling to meet the current demand and measures that have been implemented have always been to improve performance in the short time while we focus on long term acceleration of our network planning. I think we've been pretty open to say we know demand is high and we're trying to deal with your experience today, because any long term measures will take time to implement.
I have received confirmation today from our technical director that upgrades have been approved and the process is in motion. I don't have a confirmed ETA for when the new capacity will go live, but as soon as I have any rough dates I will update. He has assured me that together with MTN we are pushing Telkom for the fastest turnaround on this, and we are hoping it will be within a few weeks at the most.
We definitely know that everything is not perfect and I don't believe we've tried to say this. We know that we're struggling to meet the current demand and measures that have been implemented have always been to improve performance in the short time while we focus on long term acceleration of our network planning. I think we've been pretty open to say we know demand is high and we're trying to deal with your experience today, because any long term measures will take time to implement.
I have received confirmation today from our technical director that upgrades have been approved and the process is in motion. I don't have a confirmed ETA for when the new capacity will go live, but as soon as I have any rough dates I will update. He has assured me that together with MTN we are pushing Telkom for the fastest turnaround on this, and we are hoping it will be within a few weeks at the most.
AfriMan I understand you're in the hot seat, and to be honest I don't envy your position in the middle of all of this, but I think the enduring refrain here is that many people are frustrated because they feel (myself included) that Afrhost are not being entirely transparent about what's going on. I also understand you can't disclose everything, but if capped accounts are not being shaped, then what's been going on with p2p for the past few days? Some people have reported severely degraded performance, I have not been able to establish a connection at all, and your support guys have been less-than helpful. (although this morning p2p appears to be back to normal) It doesn't exactly instill much confidence when the telephonic support guy gives a somewhat vague response, promises to call me back after investigating, and then doesn't.
We're all paying good money for this service, and we just want some transparency and if possible, an ETA on when this will be resolved.
Exactly. A packet is a packet. Only these packets are struggling to get through certain ports right now (p2p for example) but not others.
It seems that afrihost are discriminating against certain ports in a process they like to call "optimisation". Something that the rest of the world calls "shaping".
If it were a genuine contention issue, then all packets would be lost equally. So if I can only get 600kbps in p2p, then I should get a similar speed on a speedtest (6mb/s).. Instead I get 39mb/s in a speedtest.
Afriman, do you not wonder why everyone is complaining about the same stuff in so many different threads? It's because it's a problem! Stop pretending like these are isolated incidents and asking people to pm you their details. It's insulting that you that that low of us.
Moral of the story? Other isp's work, afrihost does not! Simple.. "shaping", "optimising" etc etc.. What you call it is not the point. People aren't getting what they're paying for. Afrihost isn't delivering what they advertise. And as things stand, nobody know how long it's going to last..
If you guys admitted it from the time it became clear you cannot cope with the demand and told your customers that we are getting more capacity. LEARN FROM Previous mistakes man. Ios 7 ring a bell?
Afriman.
Explain the following to me. If P2P during a high demand period is impacted by contention as you say. Then why is it that all other protocols are running perfectly well. I was considering business 4mb and have been using a friends account while he trails another ISP and I've tested this myself. Torrent was seeded by 10 000+ and I had a rate of 17kB/s. Popped on NNTP - 53% speed from secured paid host.
Yet YouTube and Netflix within that same time frame flat lined the line at a constant 450kB/s. Contention is NOT protocol specific. It effects everything.
So a simple question again: How is the above possible.
Contention is definitely not protocol specific but it is difficult to predict. We've seen some clients experience poor realtime throughput at periods of high contention, while some seem to have poor torrents. Like exchange congestion, it's intermittent and could behave differently. I know for sure that we don't have a protocol based throttle or ruleset in place. I can't explain the end results, but I can say we don't have a ruleset in place like that.
Thanks for the update, and I'm sure there is movement to get things sorted. I do however ask this...coming from an IT Support environment myself, we monitor our systems usage and monthly report on the trends. So if we know it will take 2 months to procure additional SAN storage, we don't wait until it is over 90% utilized before starting the process. We make sure that we our average utilization is never above 80% to make provision for unforeseen growth spurts etc. (btw we do have archiving in place to avoid us just buying disks every 6 months).
Now I assume in you business you make forecasts and use history to determine these problem months combined with your client base growth. Look at previous year's posts on this forum. At around Feb-April and Aug-Nov there are always an increase in complaints usually resolved by IPC upgrades. I mean really if a software release has this much impact on your business (which happens at the same time every year!) then a few people there are not doing their work or you processes are flawed!
Again I say for more than a month, most of your clients have been complaining and taking their business elsewhere, and still not a word from Gian. I don't know what's worse, that he just doesn't care about his company's reputation or clients enough to communicate OR that if he does it will just be a version what we have been hearing from the reps for the past month with no real results.
Contention is definitely not protocol specific but it is difficult to predict. We've seen some clients experience poor realtime throughput at periods of high contention, while some seem to have poor torrents. Like exchange congestion, it's intermittent and could behave differently. I know for sure that we don't have a protocol based throttle or ruleset in place. I can't explain the end results, but I can say we don't have a ruleset in place like that.
I think looking at a network, where it's a more direct relationship to your supplier and you have other available alternatives is maybe not the best comparison. We can only order IPC from Telkom, and therefore we have to plan very carefully to ensure that we don't order too much IPC or carry too little so that we can have sustainable pricing vs overhead costs. However, the lead time to receiving upgrades is an additional complication, as this has (in the past) taken anywhere from 6 weeks to 3 months. We already had capacity upgrade plans in place, and we've had to accelerate them in the face of the unexpected increase in demand.
Gian absolutely cares, and I'm sure he won't mind me sharing that one of the first things he said after we won ISP of the Year was that we need to do better. He is very aware of the complaints, and he is a big part of why we've managed to push through our capacity order much quicker.