The other thing you have to look at is packets per second which is probably the single most important factor. For example, with the older 400MHz processors, somewhere around 22,000pps was pretty close to a hard limit. With AirMax and 64-bit packets you were limited to around 40Mbps at the AP. Bigger packets could get you up around 80Mbps. In real world scenario, 60Mbps was a pretty good average with good signal from most of the clients and 20MHz channels.
With the new processors, you should see 20-50% improvement over that. Rory -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Shawn C. Peppers Sent: Thursday, July 21, 2016 9:49 AM To: Ubiquiti Users Group Subject: Re: [Ubnt_users] Aggregate Download Capacity Real TCP on AirMax-Mx Radios I have seen a spreadsheet on the ubnt forum that does this for airfiber5x radio and dishes. You plug in the numbers for freq, dish size, etc and it provides signal, capacity, and fade margins, etc. Shawn C. Peppers Video Direct Satellite & Entertainment 866-680-8433 Toll Free 480-287-9960 Fax http://www.video-direct.tv > On Jul 21, 2016, at 11:42 AM, Dan Parrish <[email protected]> wrote: > > There are many, many reasons why there isn't just a simple X/Y chart > that shows what layer3 capacity 'Y' you can get out of your radios > given 'X' megahertz of bandwidth. > > One that we've encountered across several platforms, and really there > isn't a reason why it doesn't occur on any other platform, is layer2 > capacity being maxed while layer3 appearing to not be maxed. > > Cambium, I think recently (say in the past year), started exposing > 'frame utilization' in their PMP450 APs under statistics. It's been my > experience that in our network, we're much more likely to reach frame > utilization issues before demodulated layer 3 capacity "adds up" to > what we expect from those calculations. > > --danp > > > >> On 07/21/2016 10:06 AM, Scott Lambert wrote: >> What channel width? I would guess you can get AirRate / 2 * AMC * >> 0.90 if your -65 signals don't have -80 to -70 noise to deal with. >> >> Which band doesn't really matter. Just the channel width and >> modulation rates. >> >> 10MHz channel with 65/65 Mbps air rate / 2 = 32.5 Mbps * 80% AMC = a >> maximum of about 26 Mbps aggregate. Then take 10% off for >> engineering fudge leaves 23.4 Mbps. >> >> Unfortunately, I don't have many APs where my signals are all good >> enough to allow full modulation. >> >> Our average peak for a 10MHz channel is around 10 to 13 Mbps. >> >>> On Thu, Jul 21, 2016 at 10:26:46AM -0500, Sam Morris wrote: >>> We're looking for the amount of user traffic these radios are able >>> to pass (where these radios are APs). (The number we're trying to >>> learn is the cumulative amount of data X number of users are able to >>> concurrently pull through the AP at one time.) What is the real TCP >>> throughput on the >>> following: >>> >>> Rocket M900 >>> >>> Rocket M2 >>> >>> Rocket M5 >>> >>> I hope this isn't too ambiguous nor that it leaves too many factors >>> out of the equation (channel width for example). Please assume all >>> CPEs have >>> -65 or better so that the AP isn't struggling to resend packets to a >>> CPE with a poor connection. >>> >>> Thank you for any help you can offer. I did search the knowledge >>> base as well as the forums and didn't see a good answer. (There were >>> some good posts on the AC and AF-Xx platforms, but not for the >>> "legacy" M radios.) > > _______________________________________________ > Ubnt_users mailing list > [email protected] > http://lists.wispa.org/mailman/listinfo/ubnt_users _______________________________________________ Ubnt_users mailing list [email protected] http://lists.wispa.org/mailman/listinfo/ubnt_users _______________________________________________ Ubnt_users mailing list [email protected] http://lists.wispa.org/mailman/listinfo/ubnt_users
