Wireless Access

I'm seeing lower than expected throughtput with a controllerless setup with three IAP-215's. If I disconnect all but a single IAP-215, in close proximity with a MacBook Pro the link speed reported in mac OS is 1170mbps (MCSIndex : 9) and 802.11ac. Testing throughtput with Helios LanTest to an ethernet connected NAS shows near symetrical up/down throughput of around 13MB/s (+ or - 2MB/s). Connected via ethernet instead to the same PoE switch gives a throughput of 60MB/s.

If 13MB/s is around 104mbps, this appears to be 10x difference between the link speed and actual throughput. I this normal? I was expecting more like +40Mb/s

The IAP is on firmware 6.4.4.4-4.2.3.2_54910 with a vanilla setup and no bandwidth limits set.

On the SSID enable broadcast filter ARP in advanced.

Hi,

 

Thanks for your reply. Originally it was on 'none', then I tried 'all' (no change), then 'arp' (still no change).

 

Should I be seeing higher throughput than 13MB/s ?

I hinestly don't know anything about the Helios LAN test. How .any SSIDs are you broadcasting?

Mimo also needs an obstacle between the AP and the client for increased throughput. You also might want to have your client a minimum of 4 meters from your access point to prevent distortion that would interfere with throughput.

Hi,

SSID's - just two. This is a just a domestic application. Helios was just the first mac tool i could find. Do you know of a better tool?

If MacOS is reporting link speed and MCS index 9 wouldn't that already indicate that the physical properties, objects etc, are good i.e. If there was interference or multipath issues wouldnt affect that status?

MCS index of 9 @ 1170 correlates to the last row in the chart, in the 80mhz column, indicating 3 spatial streams.

http://www.wlanpros.com/mcs-index-802-11n-802-11ac-chart-3/

I'll try repositioning a few times and see what happens.

Try disabling the second SSID, if you can, because it does add overhead.  The "rate" is the negotiated rate, and it does not say anything realtime about throughput, aggregation inefficiency or the RF performance of any of the individual channels.  You never, ever get the speed or even half the speed of the negotiated rate, due to wireless being shared and half duplex.  There is also much more overhead involved in wireless that can decrease the rate, including your distance to the receiver and the Mimo effect.

Moving to a different location further away and, just to check it wasn't specific to that AP I used a different AP, the link speed dropped to 878mbps. Ran Helios tests which gave 13MB/s (104mbps) up and 2.3MB/s (18.4mbps) down. I've combined the data into one image below, including the AP's own quoted throughput. Afterwards I disconnected the AP and plugged the same MacBook into the same ethernet cable, and got 60MB/s (480mbps), which is perhaps limited by the NAS itself.

Here's a thread were people are seeing 40+ MB/s on consumer grade AP's. http://www.snbforums.com/threads/wifi-ac-real-life-speeds.23281/

 

 

 

Over ethernet:

 

 

Is this really normal to have such a poor throughput on current hardware? This is crazy...

It's impossible to know if it's something on the wire or environment or what. It would be far better to connect up the IAP in stand alone off a gig switch, not connected anywhere else on your network, and stand up a WPA2-PSK SSID, configure a laptop on the wired side of the switch, and run iPerf or some other simple packet analysis tool from the wireless client to the wired, taking all the rest of the network out of the equation.

 

This is not normal and in our own testing and others with clients, they are getting far far more than that, so there's likely some other factor causing the issue. 

I'm also still not clear what the test endpoints are doing. Are you just moving a large file (since you mention a LAN attached NAS server)?

 

have you tried standing up a wired IPerf server on your wire and putting iPerf on the wireless client and testing that way? File protocol plays a huge part in speeds, and for me, moving files on a MAC to an SMB-based NAS is incredibly slow, but doing iPerf or Ixia tests is near PHY rate based on SNR and MCS rate.

Many thanks - I was testing from a Mac to a wired Synology NAS but it wasnt via smb. I had thought that comparing wifi performace to a simple benchmark tool shifting 30mb files with wired performance using the AP's own ethernet cable was a valid comparison.

I will give iPerf a go and take it from there.

FWIW, file transfers of any kind are not the best measure of real/raw throughput, as file protocol overhead can really cause issues, block-ack overhead for large file blocks within the protocol, and any other mundane source of overhead (virus scanners, malware scanners, etc). Iperf and tools like it will send randomized data of a spcific format, size, protocol (TCP or UDP, etc) that removes most all the transport protocol overhead to pure TCP/UDP. 

 

iPerf 3 is the latest, iPerf2 is older. Both sides have to match.

So I managed to setup an iPerf server and on an old Dell laptop with SSD and gig ethernet. It also has an updated 802.11ac mini wifi card.

 

I have a 8 port 1gigabit PoE switch. Into that I have the IAP-215, the iPerf server laptop, and a 2015 MacBookPro with ethernet adaptop and also 3x3 wifi.

 

• The wired performance is close to the maximum, over 900Mbit/sec.
• With the Macbook on Wifi with with a link speed of 1300 (MCS index 9) iPerf reports 250-300Mbit/sec
• With the Macbook wired, and the server on Wifi, surprisingly I get a pretty similar result 250-270Mbit/sec

 

 

 

 

===========================================
MBP client (ethernet) to Dell laptop server( ethernet )  
===========================================

MacBook-Pro:documents  /Users/zzz/Downloads/iperf3 iperf -c 192.168.1.71  -p 5201 -t 10 -i 2

Connecting to host 192.168.1.71, port 5201
[  4] local 192.168.1.76 port 53725 connected to 192.168.1.71 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-2.00   sec   221 MBytes   927 Mbits/sec                  
[  4]   2.00-4.00   sec   224 MBytes   938 Mbits/sec                  
[  4]   4.00-6.00   sec   223 MBytes   937 Mbits/sec                  
[  4]   6.00-8.00   sec   223 MBytes   935 Mbits/sec                  
[  4]   8.00-10.00  sec   224 MBytes   939 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec  1.09 GBytes   935 Mbits/sec                  sender
[  4]   0.00-10.00  sec  1.09 GBytes   935 Mbits/sec                  receiver 


=======================================
MBP client (wifi) to Dell laptop server (ethernet)
=======================================

MacBook-Pro:documents   /Users/zzz/Downloads/iperf3 iperf -c 192.168.1.71  -p 5201 -t 10 -i 2

Connecting to host 192.168.1.71, port 5201
[  4] local 192.168.1.66 port 53756 connected to 192.168.1.71 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-2.00   sec  61.6 MBytes   258 Mbits/sec                  
[  4]   2.00-4.00   sec  67.7 MBytes   284 Mbits/sec                  
[  4]   4.00-6.00   sec  71.7 MBytes   301 Mbits/sec                  
[  4]   6.00-8.00   sec  70.3 MBytes   295 Mbits/sec                  
[  4]   8.00-10.00  sec  71.8 MBytes   301 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   343 MBytes   288 Mbits/sec                  sender
[  4]   0.00-10.00  sec   343 MBytes   288 Mbits/sec                  receiver


=========================================
MBP client (ethernet) to Dell laptop server (Wifi)
=========================================

MacBook-Pro:documents  /Users/zzz/Downloads/iperf3 iperf -c 192.168.1.73  -p 5201 -t 10 -i 2

Connecting to host 192.168.1.73, port 5201
[  4] local 192.168.1.76 port 53925 connected to 192.168.1.73 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-2.00   sec  60.4 MBytes   253 Mbits/sec                  
[  4]   2.00-4.00   sec  63.9 MBytes   268 Mbits/sec                  
[  4]   4.00-6.00   sec  62.4 MBytes   262 Mbits/sec                  
[  4]   6.00-8.00   sec  60.3 MBytes   253 Mbits/sec                  
[  4]   8.00-10.00  sec  56.8 MBytes   238 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   304 MBytes   255 Mbits/sec                  sender
[  4]   0.00-10.00  sec   304 MBytes   255 Mbits/sec                  receiver


*************

 

You should try to use the option -w512k so that you use a larger window size..

 

More info here:  http://blog.softlayer.com/2011/using-iperf-to-troubleshoot-speedthroughput-issues/

TO add to Colin, you should use 4-8 streams (use '-P 4' or '-P 8'). 

I'll update with the throughputs with the modified iPerf command shortly, but menawhile, I found this article which mentions a 64k window size on OSX that cripples performance on SMB and AFP shares.

 

http://www.anandtech.com/show/7085/the-2013-macbook-air-review-13inch/10

 

 

 

The highest numbers I saw were with 512k, 4 clients at a sum total of +405mbps. A single client test saw 359mbps.

 

Are these numbers as expected over a link reporting 1300 (MCS index:9) ?

 

 

 

MacBook-Pro:documents /Users/zzz/Downloads/iperf3 iperf -c 192.168.1.71  -p 5201 -t 10 -i 10 -w 512k -P 4
Connecting to host 192.168.1.71, port 5201
[  4] local 192.168.1.66 port 50761 connected to 192.168.1.71 port 5201
[  6] local 192.168.1.66 port 50762 connected to 192.168.1.71 port 5201
[  8] local 192.168.1.66 port 50763 connected to 192.168.1.71 port 5201
[ 10] local 192.168.1.66 port 50764 connected to 192.168.1.71 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   115 MBytes  96.5 Mbits/sec                  
[  6]   0.00-10.00  sec   114 MBytes  95.3 Mbits/sec                  
[  8]   0.00-10.00  sec   120 MBytes   101 Mbits/sec                  
[ 10]   0.00-10.00  sec   138 MBytes   116 Mbits/sec                  
[SUM]   0.00-10.00  sec   487 MBytes   408 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   115 MBytes  96.5 Mbits/sec                  sender
[  4]   0.00-10.00  sec   115 MBytes  96.4 Mbits/sec                  receiver
[  6]   0.00-10.00  sec   114 MBytes  95.3 Mbits/sec                  sender
[  6]   0.00-10.00  sec   113 MBytes  94.9 Mbits/sec                  receiver
[  8]   0.00-10.00  sec   120 MBytes   101 Mbits/sec                  sender
[  8]   0.00-10.00  sec   120 MBytes   100 Mbits/sec                  receiver
[ 10]   0.00-10.00  sec   138 MBytes   116 Mbits/sec                  sender
[ 10]   0.00-10.00  sec   138 MBytes   116 Mbits/sec                  receiver
[SUM]   0.00-10.00  sec   487 MBytes   408 Mbits/sec                  sender
[SUM]   0.00-10.00  sec   486 MBytes   407 Mbits/sec                  receiver

iperf Done.

 

MacBook-Pro:documents  /Users/zzz/Downloads/iperf3 iperf -c 192.168.1.71  -p 5201 -t 10 -i 10 -w 512k 
Connecting to host 192.168.1.71, port 5201
[  4] local 192.168.1.66 port 50868 connected to 192.168.1.71 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   428 MBytes   359 Mbits/sec                  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   428 MBytes   359 Mbits/sec                  sender
[  4]   0.00-10.00  sec   428 MBytes   359 Mbits/sec                  receiver

iperf Done.

That could be about right, depending on how clear the channel is and how silent your host client(s) are (if your test laptops are or have any other software exchaning data with the network, it can impact the numbers). But that's about right, a good real-world number I think (it's ballpark). Note that the connected rate is the phy rate, which is half-duplex.