How Well Do AC Routers Handle Mixed Networks?

[BigGuy]

Thanks for the interesting article.

After reading it, I was left with this question:

How do the results with simultaneous transfers with "mixed-standard" devices compare with the throughput of simultaneous "single-standard" transfers (e.g. AC867 + AC867 or AC867 + AC433)?

I would expect rates to drop in the above cases, but I'm not sure how much more or less the drop would be compared to the the cases you tested.

Another way to look at your non-capped test results from tables 2 & 3 is to look at how the total data rate is affected by the various combinations of devices:

1 Client: AC867 : 199.4 Mbps
(Assume this corresponds to max possible wireless data rate through the router for the test configuration used.)

2 Clients: AC867+N450 : 130+59 = 189 Mbps
loss: 199.4 - 189 = 10.4 Mbps
10.4/199.4 = 10% loss in total data rate

2 Clients: AC867+N150 : 120+12.8 = 142.8 Mbps
loss: 199.4 - 142.8 = 56.6 Mbps Mbps
56.6/199.4 = 28.7% loss in total data rate

2 Clients: AC867+A54 : 140+6.2 = 146.2 Mbps
loss: 199.4 - 146.2 = 53.2 Mbps
53.2/199.4 = 26.6% loss in total data rate

2 Clients: AC867+AC867 : ???
(will there be 0%, 10% or 25% loss?)

2 Clients: AC867+AC433 : ???
(will there be 0%, 10% or 25% loss?)

4 Clients: AC867+N450+N150+A54: 57.3+29.5+12.0+2.1 = 100.9 Mbps
loss: 199.4 - 100.9 = 98.5 Mbps
98.5/199.4 = 49% loss in total data rate.

4 Clients: AC867+AC867+AC867+AC867: ???
(will there be 0%, 10%, or 50% loss?)

 

[thiggins]

If all clients use the same standard, bandwidth sharing should be equal... theoretically. In the real world, differences in adapter design, host platform OS and architecture and RF environment result in some unequal sharing.

Running multiple IxChariot streams on a single AC client usually results in higher throughput than from a single stream. I ran this test on early AC router reviews. The attached plot shows up to 500 Mbps bandwidth available. But this is with a 3x3 client. I used only 2x2 in the article.

I ran a quick check with two NETGEAR A6200 (2x2) clients that I didn't publish in the article. Bandwidth sharing was quite unequal as shown in mixed_ac_two_2x2_clients.jpg (attached) and didn't result in higher bandwidth. One of the two clients alone got 181 Mbps and the other was 168 Mbps.

Sounds like a follow up article would be of interest.

 

[sfx2000]

Tim,

Good stuff... so it does look like 11ac is better behaved than 11n was in 2.4Ghz space.

Goes without saying though, the original concept of add vs. replace is still valid if one has the space and spectrum/channels to do it.

I had an error in an earlier posting on this thread - for 11n in 5Ghz space, it looks like most of the 11ac chipsets are three stream there, so apologies if this caused any confusion.

Perhaps at some point, we might see some of the 11ac benefits trickle down into the 2.4Ghz space with updated 11n drivers - the CCA (Clear Channel Assessment) feature alone could make life easier for many - wide channels then could be a realistic option for some folks. There's other features in 11ac that could be backported into 11n without breaking legacy as well.

Wonder if we'll see a point release/bug fix for 11n once work is completed on 11ac. Wouldn't be the first time that IEEE 802 has issued a corrigendum - happened earlier with 802.16 (Wimax).

http://grouper.ieee.org/groups/802/11/Reports/802.11_Timelines.htm

 

[thiggins]

Follow up article

How Much Throughput Can You Really Get From An AC Router?

 

[L&LD]

The reason why we need AC Wave 2 routers with 4+ Antenna

How Much Throughput Can You Really Get From An AC Router?

Thank you for such a quick follow up!

Could this indicate what we might have when Wave 2 AC becomes available (clients and routers, of course) and we have 4 antenna or more and MU-MIMO?

I was actually surprised that the total bandwidth couldn't be exceeded with multiple (same class) clients. I thought that the 'hard limit' would be closer to that indicated by your WAN/LAN, LAN/WAN and Total Simultaneous tests.

Learning, learning - what a great site!

 

[offroad]

What are some good tools to use to test bandwidth throughput? Am fairly new to this and just using speedtest.net with clients. But thinking there have got to be local LAN tools that can be used.

 

[thiggins]

What are some good tools to use to test bandwidth throughput? Am fairly new to this and just using speedtest.net with clients. But thinking there have got to be local LAN tools that can be used.

http://www.smallnetbuilder.com/lanwan/lanwan-basics/32252-three-more-ways-to-measure-network-speed