I'm pondering some things about how to consider the capacity of an 802.11n wireless multipoint access sector and would like some feedback from the other forum gurus on the subject.

Suppose a sector runs 2x2 802.11n on a 20Mhz channel, and for the sake of simplicity, let's assume that all clients achieve full modulation rate of 130x130.
In my experience, this tends to cap off in real throughput at around 60Mbps total throughput (tx and rx speeds combined)

Does this seem reasonable to you?

Now the main question:
802.11n is not a timing-dependent technology where x amount of time is allocated for upstream and y amount allocated for downstream, so is it reasonable to consider the saturation of a sector based on just the total bandwidth permitted to each customer, regardless of direction?

i.e. - Would there be any difference between selling 8 links @ 5x2 vs selling 8 links @ 2x5 where capacity planning is concerned?

As a WISP that is 100% Mikrotik, sector capacity planning for us has become a two horse race Mimosa or Cambium, Mikrotik and UBNT are not in the same league, as a long term supporter of Mikrotik it is sad they have not given wireless the support that it deserved for high volume purchasers like WISP's, only now has sync become available, Mimosa AP to Mikrotik CPE in 802.11 mode does not suffer from disconnections in high noise locations unlike MT AP to MT CPE that disconnect regularly and have to use NV2 , also Mimosa AP has a much higher throughput in that mode, Cambium are actively developing ePMP Elevate solution which when ready will result in Cambium AP to Mikrotik CPE, unfortunately the future for Mikrotik wireless is not so bright?

I'm not looking to select gear for wireless shots - I'm just trying to (in)validate my thoughts as to the effect of bandwidth allocation on sector capacity.
I.e. - does giving a customer more upstream detract more from a sector than giving them the same amount of additional downstream would?

ZeroByte,

I don't have the knowledge about this subject to even talk about the technical details with you. However, hopefully I can bring out a good thought process for you to consider.

Mission:
You're not selling bandwidth, you're selling user experience. You're selling user happiness. Give your customer's a fabulous experience with your service, and they'll pay the amount that is fair for the both of you. You'll both be happy.

Idea:
My understanding: with 802.11n, the number of devices that are holding onto the rights to transfer bits effects the experience for others who are waiting their turn to do the same thing. The best thing you can do is to get people out of these queues as fast a possible. If a client opens a Pinterest page, they're not "downloading" in their minds. Yet, they've got a lock on a WiFi channel for the next 20MBs!

Instead of throttling them, give it to them as fast as you can so that another user will not experience delays.

Selling:
How you go about selling this might state things like: the density of the network, number of people online at a given moment, or restrictions on "downloads" (if it looks like that), but the browsing experience is something you need to tune. Limiting "bandwidth" on a WiFi service is just a way to cause a ton of devices to all fight for airtime - potentially slowing things down for everyone. Let people know that "if we have the capacity, we share it with you, our wonderful customer!"

My apologies if that makes no sense in the real world.

I'm not planning for end-user WiFi here.

I'm looking for technical information here, not product delivery paradigms.

Now the main question:
802.11n is not a timing-dependent technology where x amount of time is allocated for upstream and y amount allocated for downstream, so is it reasonable to consider the saturation of a sector based on just the total bandwidth permitted to each customer, regardless of direction?

i.e. - Would there be any difference between selling 8 links @ 5x2 vs selling 8 links @ 2x5 where capacity planning is concerned?

If you question is based on how much upload/download speeds achievable to X amount of customers, then location (single AP vs Multiple AP's on site) will determine what is achievable
and will also fluctuate?

This is outdoor fixed wireless backhaul.

If I have a customer who adds X amount to their allowed upstream on the backhaul circuit (i.e. there's a layer3 break between the customer's actual LAN and this wireless link) and I allow 3 more Mbps on their upstream-facing queue, and they start using it, will that have a worse impact on the sector's overall capacity than if I were to add 3Mbps to their download queue and they start using that?

This assumes a good sector with all clients at max modulation rate, etc.

This is outdoor fixed wireless backhaul.

If I have a customer who adds X amount to their allowed upstream on the backhaul circuit (i.e. there's a layer3 break between the customer's actual LAN and this wireless link) and I allow 3 more Mbps on their upstream-facing queue, and they start using it, will that have a worse impact on the sector's overall capacity than if I were to add 3Mbps to their download queue and they start using that?

This assumes a good sector with all clients at max modulation rate, etc.

Hello ZeroByte,

If we're talking about pure 802.11n, or nstreme or nv2 with download ratio set to dynamic then not, it should have the same cost for sector's overall capacity regardless of direction of the traffic.
However, since there is almost no ideal example of that in real world and ROS has proprietary wlan driver I think that only someone from Mikrotik can give definitive theoretical answer.


In real world, my experience with the matter is that in the essence we're dealing with the time allocation of the sector (and with the lacking of the fine manual control over it in ROS).

Time of the sector is being divided for downstream and upstream and additionally per each customer.
So, if you allow someone to have higher, let say, upstream speed then the sector will have to allocate higher percentage of its time on that user in upstream direction.

As we're talking about half-duplex medium, that obviously means that the sector will have less time/capacity left for other connected costumers in both directions combined.

The same thing is with the higher downstream speed, sector just must allocate more percentage of its time for that costumer.

Regards,
M.