Afrihost Uncapped ADSL Feedback (Pt3)

[AfriMan]

No clue, The east got a full IPC upgrade and our shaping only eases off after 11pm.


Afriguy, just to give you an illustration of my lack of faith in your product right now. I woke up at 4am this morning to catch up on some work. I actually thought let me just check if I'm unshaped. ( and yes I was) The fact that I considered being shaped and wouldn't have been surprised to see shaping at 4am illustrates how much I think your letting us the users and your company down.

Even though we received a full upgrade, we are still going to proceed cautiously to ensure that we maximise the potential of the capacity we've secured. What we've seen in the past is that we've opened up the network completely and that capacity is consumed very quickly and it's then very difficult to ensure good performance on priority services because contention is driven up. That's also unfair in some ways, and we'd prefer to proceed in a way that we can deliver realistic expectations of the network's performance but also be able to guarantee good realtime uncapped performance at all times, which is a cornerstone of our shaping/network management model.

I know that it's not inspiring a lot of confidence, but we do believe that this is the best way to proceed, and I don't believe that this shaping will continue for much longer. The difficulty is that it's very difficult to gauge realtime demand because many businesses and users are on holidays because of all the public holidays this week and next week. We doing our best to gather all the metrics and reports we need to find the best balance for our Uncapped users.

 

[AfriMan]

But premium performance on realtime services has not been maintained as streaming and youtube are both still performing very poorly, as I mentioned in my post earlier.

That should definitely not be the case, becuase if we're maintaining shaping, it means we have the capacity to ensure those services are not contended. Not sure if you have PM'd AfriGuy yet, but please PM him or me ASAP so we can do some troubleshooting ASAP on your line and account.

 

[HapticSimian]

Here's a summation of my frustration: I've seen reasonable shaping once as far as memory serves. Every other time I've bothered to check - which must surely be a few 100 times over the months - it's been at the magical 18.8/6.25 mark. That is inexcusable, all the more so when taking into account the times of day we're talking.

I've never been one to run my line at capacity for days, or even hours, on end. I have uncapped because I don't want to have the bother of monitoring usage. And don't give me the 'not the product for you' rationalisation; it was. For years. YOU, Afrihost, changed it unilaterally. As for your Turbocharge-nonsense: I'm as interested in watching a clock as what I am in watching overall usage, which is not at all. When I want or need a download of a couple of 100 megs I do not want to be jumping through hoops to get it in a reasonable time frame. Trickling in at a 5th my line speed isn't reasonable.

 

[TwilightWalker]

We can't really use hours as a guideline, as there is no way to predict demand. So while we can have some idea of when people leave work or go to bed, demand can sometimes continue well into the early hours of the morning.

Remember that we are also only releasing new capacity very slowly, to ensure that we maintain premium performance on realtime services and low latency, and the determine how much we can unshape so that we have a good balance until we receive additional capacity.

But going forward we'll continue to shape according to realtime demand and not time based schedules.

That's fair but maybe saying things like "off peak" and "after hours" can be misinterpreted. It is just a bit hard to believe that the network is as busy at 11AM as it is at 7PM and 11PM.

 

[AfriMan]

Here's a summation of my frustration: I've seen reasonable shaping once as far as memory serves. Every other time I've bothered to check - which must surely be a few 100 times over the months - it's been at the magical 18.8/6.25 mark. That is inexcusable, all the more so when taking into account the times of day we're talking.

I've never been one to run my line at capacity for days, or even hours, on end. I have uncapped because I don't want to have the bother of monitoring usage. And don't give me the 'not the product for you' rationalisation; it was. For years. YOU, Afrihost, changed it unilaterally. As for your Turbocharge-nonsense: I'm as interested in watching a clock as what I am in watching overall usage, which is not at all. When I want or need a download of a couple of 100 megs I do not want to be jumping through hoops to get it in a reasonable time frame. Trickling in at a 5th my line speed isn't reasonable.

The only time we changed the product was when we moved networks from IS to MTN, when we did away with throttling in favour of realtime shaping based on demand. The product has never changed fundamentally, only the variables surrounding available capacity and us trying to ensure the best network performance based on the resources we have. That is the product, so you kind of do need to make a decision on whether this is the right product for you. Uncapped on our network for the foreseeable future is going to remain demand driven, with features like Turbocharge to provide unshaped relief as and when clients want that.

I've also asked several times for suggestions on how we can improve the product, and I promise that if there are any reasonable, actionable suggestions - I will make sure they reach management.

 

[AfriMan]

That's fair but maybe saying things like "off peak" and "after hours" can be misinterpreted. It is just a bit hard to believe that the network is as busy at 11AM as it is at 7PM and 11PM.

We're seeing pretty consistent demand - remember that many people will schedule P2P services and just leave them running while they go to bed. So while realtime services may taper off, we still see a huge demand for P2P and NNTP services.

 

[HapticSimian]

The only time we changed the product was when we moved networks from IS to MTN, when we did away with throttling in favour of realtime shaping based on demand. The product has never changed fundamentally, only the variables surrounding available capacity and us trying to ensure the best network performance based on the resources we have. That is the product, so you kind of do need to make a decision on whether this is the right product for you. Uncapped on our network for the foreseeable future is going to remain demand driven, with features like Turbocharge to provide unshaped relief as and when clients want that.

I've also asked several times for suggestions on how we can improve the product, and I promise that if there are any reasonable, actionable suggestions - I will make sure they reach management.

OF COURSE it has changed fundamentally - it went to working - and working extremely well at that - for a segment of your clients one day, to being a constant source of frustration the next. The hows and whys, from an end-user's perspective, are wholly irrelevant to you materially changing your offering, and being on a half-year long damage limitation drive since.

It's not your customers' responsibility to suggest fixes.

 

[AfriMan]

OF COURSE it has changed fundamentally - it went to working - and working extremely well at that - for a segment of your clients one day, to being a constant source of frustration the next. The hows and whys, from an end-user's perspective, are wholly irrelevant to you materially changing your offering, and being on a half-year long damage limitation drive since.

It's not your customers' responsibility to suggest fixes.

If it's a demand based product, the variables are the demand and resources, not the workings of the product. In essence you are saying we should change the product, which we can't do. The design and workings of the product have not been altered.

Again, I truly believe that the forums are here for us to arrive at better solutions together. If you're not willing to make suggestions, then you need to trust us to do what we feel is best weighing up the various factors. I'm also trying to give you insight into what those factor sare, so I hope that is what is coming across.

 

[Chevron]

Just out of curiosity how many turbos do you guys get a month and how much does it cost for more?

 

[AfriMan]

Just out of curiousity how many turbos do you guys get a month and how much does it cost for more?

You get 300 Free minutes per month, and then you can purchase in increments of R29. The number of minutes per topup depends on your line speed - so 10Mbps topup buys you 60 minutes, where a 4Mbps topup buys you 120 (if I'm not mistaken)

 

[Chevron]

You get 300 Free minutes per month, and then you can purchase in increments of R29. The number of minutes per topup depends on your line speed - so 10Mbps topup buys you 60 minutes, where a 4Mbps topup buys you 120 (if I'm not mistaken)

Cool thanks. I'm just looking at all the info regarding all accounts and trying to see if I have some awesome idea to help everyone.

 

[AfriMan]

Cool thanks. I'm just looking at all the info regarding all accounts and trying to see if I have some awesome idea to help everyone.

All ideas are definitely welcome. I take anything that's workable straight to management - promise

 

[HapticSimian]

If it's a demand based product, the variables are the demand and resources, not the workings of the product. In essence you are saying we should change the product, which we can't do. The design and workings of the product have not been altered.

Again, I truly believe that the forums are here for us to arrive at better solutions together. If you're not willing to make suggestions, then you need to trust us to do what we feel is best weighing up the various factors. I'm also trying to give you insight into what those factor sare, so I hope that is what is coming across.

I feel an analogy coming on:

Let's say you lease a fully-functioning car from me, and happily drive it daily - to work and back, to the shops, even on holiday - for 18 months. In fact you're so happy with your car that you suggest my services to anyone you discuss the topic of transport with.

One bright sunshiny day I come along with a spanner & I remove the back seat, so now you can't transport more than one passenger any more. Except for the times when I, seemingly randomly, give you a back seat for a bit before just snapping it away again without a moment's notice. At least I give you the option of renting a back seat from me for hour-long intervals...

No matter how many good, solid reasons I have for taking your back seat, would you be satisfied? Would me saying I've not fundamentally changed your car make you feel better?

 

[AfriMan]

I feel an analogy coming on:

Let's say you lease a fully-functioning car from me, and happily drive it daily - to work and back, to the shops, even on holiday - for 18 months. In fact you're so happy with your car that you suggest my services to anyone you discuss the topic of transport with.

One bright sunshiny day I come along with a spanner & I remove the back seat, so now you can't transport more than one passenger any more. Except for the times when I, seemingly randomly, give you a back seat for a bit before just snapping it away again without a moment's notice. At least I give you the option of renting a back seat from me for hour-long intervals...

No matter how many good, solid reasons I have for taking your back seat, would you be satisfied? Would me saying I've not fundamentally changed your car make you feel better?

I don't see how a car can be compared to a demand driven product? It's a fixed product with guaranteed working parts and components, no variables at all. So probably not a good analogy to really work with, in my opinion.

But comparisons aside, as I've said, we're open to suggestions. If you think you have an idea on how we can make the product better, I'm here to listen, and I promise if it's something that can be actioned, we'll look at it.

 

[aybbleek]

Just out of curiosity how many turbos do you guys get a month and how much does it cost for more?

300 turbos a month, think its like R29 for 120 minutes there after?

 

[Chevron]

All ideas are definitely welcome. I take anything that's workable straight to management - promise

Not sure how workable this is, but here goes. This should also alleviate the headache of managing turbos.

So my thinking process.

First I listed the home accounts speeds and pricing, then did the same for business and worked out the price increase in % terms and listed the rand value.

So it looks like this:

Home
2meg R297
4Meg R397
10meg R797

Biz
2meg R397 33% R100
4Meg R697 75% R300
10meg R997 25% R200

The reason I this is is to see how far away the home guys are from 100% unshaped.

So my idea is to offer a bolt on where you remove the turbos, but offer a guaranteed minimum shaping value on downloads/p2p for a nominal fee.

eg.

The 2meg that's currently on 18%/6.25% can pay an extra R50 a month and will never get shaped worse than 55%/45% ever on his account.

4meg would pay R150 for this privilege and the 10meg guys R100.

 

[Bl1zz4rd]

I've also asked several times for suggestions on how we can improve the product, and I promise that if there are any reasonable, actionable suggestions - I will make sure they reach management.

Someone suggested, a while back in this thread, a QoS feature being placed in the ClientZone. I know that this is a huge task to implement, but if done properly, I think it could work very well and solve a lot of people's problems. I know it'd solve mine.

If we could, via the ClientZone, choose which services we wanted to give priority to at any given moment, we could in essence be in control of our own data. This is quite a nice ideal to move towards.

There could be buttons for prioritizing realtime, P2P and downloads and, for instance, when we choose to prioritize our P2P, if we want a download quickly, other services are shaped heavily (like the heavy shaping users are experiencing now for P2P), so then browsing would be slow, but it'd be our own doing. As soon as we wanted to change back, we could just press another button, and we could do so. Basically, users could just switch based on the usage they know they'll be doing, giving themselves the best experience tailored to their own specific needs.

I'm not sure just how possible that all is, but I genuinely believe that it should at least go through to management for consideration.

 

[Qrox]

Great... 90 kB/s for downloading a Uplay patch...
So sick and tired of the shaping. We are paying for a 4mbs connection not a less than 1mbs....

 

[Mualliq]

Great... 90 kB/s for downloading a Uplay patch...
So sick and tired of the shaping. We are paying for a 4mbs connection not a less than 1mbs....

I dream of the days I could get 90 kbps, and I have a 10 mbps line. I get 20 kbps p2p for 16-18 hours a day. That is a fiftieth of potential line speed, which means that afrihost has fifty people competing for every byte that one person previously got, or that that they are 5000% oversubscribed.

 

[Chevron]

Someone suggested, a while back in this thread, a QoS feature being placed in the ClientZone. I know that this is a huge task to implement, but if done properly, I think it could work very well and solve a lot of people's problems. I know it'd solve mine.

If we could, via the ClientZone, choose which services we wanted to give priority to at any given moment, we could in essence be in control of our own data. This is quite a nice ideal to move towards.

There could be buttons for prioritizing realtime, P2P and downloads and, for instance, when we choose to prioritize our P2P, if we want a download quickly, other services are shaped heavily (like the heavy shaping users are experiencing now for P2P), so then browsing would be slow, but it'd be our own doing. As soon as we wanted to change back, we could just press another button, and we could do so. Basically, users could just switch based on the usage they know they'll be doing, giving themselves the best experience tailored to their own specific needs.

I'm not sure just how possible that all is, but I genuinely believe that it should at least go through to management for consideration.

I think you're missing the point of network shaping. p2p kills networks. that's why it's shaped. If you gave people the option to unlimitedly prioritize p2p then the network would die. If you wanna download some p2p quickly that's what the turbo boosts are for.

The business accounts have protocol prioritization, but then that's why they pay more.