in network design maybe it is quoted as he said.

But I think the reason the lowest value is used is to give the provider some flexibility.

As things do work better when larger shared pipes are used even with the same contention ratio, so the provider I mentioned with the shared gbit pipes may in future change them to 10gbit pipes and stick 1000 customers on instead of 100. If they were quoting 100:10 ratio's to customers tho that would mean they would need to change what they selling whilst quoting a simple 10:1 they dont, they just make the network switch.

Edited by Chrysalis (Sat 19-Jul-14 15:25:21)

In reply to a post by RobertoS:

Given the identity of the person who avers such figures are used, I tend to believe him. What I suggested was where I thought he may encounter it, but it was just a guess. It is certainly in some arcane environment.

Purely as a ratio it is the same thing but as an implementation it definitely isn't. 50x1Mb/s users on 1Mb/s backhaul will not experience the same service level as 500x1Mb/s users on 10Mb/s backhaul.

What I don't know (and didn't say anything about) is if/when anyone chooses to quote the implementation figures rather than simply giving the ratio. If we assume that no sensible network designer would ever go with a genuine ':1' ratio then the difference does become rather moot.

This does make good sense. Very good way to explain it.

Purely as a ratio it is the same thing but as an implementation it definitely isn't. 50x1Mb/s users on 1Mb/s backhaul will not experience the same service level as 500x1Mb/s users on 10Mb/s bac

Yep it's a very different situation. That being said I haven't seen ISPs using these kind of ratios when selling broadband for a long time now so perhaps it's not that important to the end user nowadays to concern themselves with the logic.

I think we are agreed that the first line of the OP, which is where this started, is drivel by the sales/support person who said it. I wouldn't be surprised if it isn't still in the scripts .

In reply to a post by Andrue:

Purely as a ratio it is the same thing but as an implementation it definitely isn't. 50x1Mb/s users on 1Mb/s backhaul will not experience the same service level as 500x1Mb/s users on 10Mb/s backhaul.

What I don't know (and didn't say anything about) is if/when anyone chooses to quote the implementation figures rather than simply giving the ratio. If we assume that no sensible network designer would ever go with a genuine ':1' ratio then the difference does become rather moot.

Capacity planning doesn't work on ratios in my experience it works on expected peak utilisation per customer.

In reply to a post by ukhardy07:

My logic is 50:1 is the same as 500:10
There is equal likelihood in both cases
50:1 is a 2% chance

500:10 is still a 2% chance

Similar 5000:100 is still the same probability.

Not actually true. Statistical contention means it's less likely that 10 users out of 500 are maxing out their connections than 1 user out of 50.

Virgin Media are a case in point. They allocate tons of bandwidth to each customer compared with many but because it takes so few customers saturating their bandwidth to cause issues they are way more prone to issues than LLU operators or whomever.

This puts it beautifully.

But isn't the difference between fixed speed and rate adaptive being forgotten in parts of this discussion? (As I think I've mentioned before, unless I scrapped it at the review stage).

In reply to a post by RobertoS:

But isn't the difference between fixed speed and rate adaptive being forgotten in parts of this discussion? (As I think I've mentioned before, unless I scrapped it at the review stage).

No idea. Was just answering that one specific point about how, say, 500Mb of capacity sold on a 10Mb pipe is preferably to 50Mb sold on a 1Mb pipe.

Rate adaption is why, amongst other reasons, as I mentioned in a different post capacity planning is done in kbps per customer at peak utilisation time rather than based around contention ratios.

In reply to a post by Ignitionnet:

Rate adaption is why, amongst other reasons, as I mentioned in a different post capacity planning is done in kbps per customer at peak utilisation time rather than based around contention ratios.

Quite . That's the whole point of the thread, as opposed to the off-topic bit.