Wireless Access

Hi All,

I've a customer setup a mesh link between 2 IAP-175 which are about 400 meters apart with clear LOS. Both IAPs are attached with an Ant 2x2 5614-U external antenna.

 

However we noticed that LED indictors on both did not go beyond 2 bars and the RSSI value always hovers around 35 ~ 40 dbm only. We ran an iperf throughput tests with a laptop connected to mesh point and mesh portal at the other end. The results showed that from the radios stats, the packets are transmitting at around 270Mbps and receiving at 300Mbps which is reasonable, however the channel utilisation is hitting only 39% and the throughput is only 80Mbps. This is very far off from the thoretical max of 150Mbps.

 

Primary-Point# show ap mesh ni eighbours

Neighbor list
-------------
MAC                Portal  Channel  Age  Hops  Cost  Relation          Flags  RSSI  Rate Tx/Rx  A-Req  A-Resp  A-Fail  HT-Details       Cluster ID
---                ------  -------  ---  ----  ----  --------          -----  ----  ----------  -----  ------  ------  ----------       ----------
xxxxxxxxxx  Yes     149+     0    0     4.00  P 5d:15h:41m:23s  HLK    39    300/300     1      1       0       HT-40MHzsgi-2ss  194c96b88cb7586fe383f80f89d96cc

Total count: 1, Children: 0

 

Tx HT 216 Mbps                 72720
Tx HT 240 Mbps                 281628
Tx HT 243 Mbps                 500425
Tx HT 270 Mbps                 37120968  <<<<<<<<<<<<<

 

Rx HT 240 Mbps    1215691
Rx HT 243 Mbps    371076
Rx HT 270 Mbps    82574913
Rx HT 300 Mbps    3144872160  <<<<<<<<<<<<

 

Channel Busy 1s                39   <<<<<<<<<<<<<<<<
Channel Busy 4s                43
Channel Busy 64s               39
Ch Busy perct @ beacon intvl   44 42 43 44 44 42 42 44 45 19 25 46 41 45 42 43 45 43 45 44 45 43 42 46 45 42 46 42 47 43
Rx Time perct @ beacon intvl   36 33 35 36 35 35 34 36 36 16 20 38 33 37 34 34 35 35 37 35 34 35 34 36 35 33 37 34 38 34

 

The iperf parameters we used at both sides are as follows :

 

Iperf –c 1.2.3.4   -w 512k  -t 60   -P 4

Iperf –s  -w 512k

 

Initially I suspect that there's some problem with the wifi NIC of the clients or some bottlenecks along the network path. But we did the same test with both same clients on the wired network and they are able to achieve > 900Mbps thoughput. So there should not be any issue with the machines.

 

Please advice how we can bump up the channel utilisation and throughput.

 

Thanks

 

Hi All,

 

Any advice is much appreciated.

 

 

Thanks! :manembarrassed:

 

What type of wireless nic are you using ? 3x3:3 or 2x2:2 ?

 

Depending on the type of nic you are using it will determine speeds you can obtain too, also remember that you will be using 802.11gn (20 Mhz) for client traffic when meshing .

 

Why are you trying to accomplish by increasing the utilization on that channel ? 

 

Can you please share the show ap mesh topology ?

 

 

 

 

Hi Fabian,

 

Thanks for the reply. I do not have the "show ap mesh topo" output. What i have are only the "show ap mesh neighbour" and "show ap mesh link:

 

Primary-Portal# show ap mesh link

Neighbor list
-------------
MAC                Portal             Channel  Age  Hops  Cost  Relation         Flags  RSSI  Rate Tx/Rx  A-Req  A-Resp  A-Fail  HT-Details       Cluster ID
---                ------             -------  ---  ----  ----  --------         -----  ----  ----------  -----  ------  ------  ----------       ----------
x:x:x:x  x:x:x:x  149+     0    1     7.00  C 5d:15h:38m:4s  HLK    37    300/300     1      1       0       HT-40MHzsgi-2ss  194c96b88cb7586fe383f80f89d96cc

Total count: 1, Children: 1
Relation: P = Parent; C = Child; N = Neighbor; B = Blacklisted-neighbor
Flags: R = Recovery-mode; S = Sub-threshold link; D = Reselection backoff; F = Auth-failure; H = High Throughput; L = Legacy allowed
 K = Connected; U = Upgrading; G = Descendant-upgrading; Z = Config pending; Y = Assoc-resp/Auth pending
 a = SAE Accepted; b = SAE Blacklisted-neighbour; e = SAE Enabled; u = portal-unreachable; o = opensystem

Primary-Portal# show ap mesh neighbours

Neighbor list
-------------
MAC                Portal             Channel  Age  Hops  Cost  Relation          Flags  RSSI  Rate Tx/Rx  A-Req  A-Resp  A-Fail  HT-Details       Cluster ID
---                ------             -------  ---  ----  ----  --------          -----  ----  ----------  -----  ------  ------  ----------       ----------
x:x:x:x  x:x:x:x  149+     0    1     7.00  C 5d:15h:37m:45s  HLK    38    300/300     1      1       0       HT-40MHzsgi-2ss  194c96b88cb7586fe383f80f89d96cc

Total count: 1, Children: 1
Relation: P = Parent; C = Child; N = Neighbor; B = Blacklisted-neighbor
Flags: R = Recovery-mode; S = Sub-threshold link; D = Reselection backoff; F = Auth-failure; H = High Throughput; L = Legacy allowed
 K = Connected; U = Upgrading; G = Descendant-upgrading; Z = Config pending; Y = Assoc-resp/Auth pending
 a = SAE Accepted; b = SAE Blacklisted-neighbour; e = SAE Enabled; u = portal-unreachable; o = opensystem

 

But as you can see above, the mesh link is using 40Mhz channels. How can I know whether the wireless nic is using 2x2 or 3x3? Any CLI command to show that?

 

Thanks

The link between the mesh portal and mesh point is using a 40 MHz on 802.11an but when a device connects it will use 802.11bgn.

Are using a windows or a Mac device?

We are using both windows laptops to perform the iperf test.

Wolfie99,

 

That is at 400 Meters?  Those are great numbers.  The channel is 39% busy which would definitely limit your throughput.

 

What do you think it should be?

 

Hi Joseph,

Thanks for the reply. But I thought the 39% channel busyness was due to the packets traversing the link when I'm performing the iperf test?

I've actually checked with one of your local TAC and he informed me that he managed to achieve a throughput of 140+ Mbps with a channel util rate of 80+% in a lab environment.

 

Is it possible to achieve that rate at a distance of 400m apart for the IAPs? or is it a limitation?

 

 

Thanks and Regards,

 

Wolfi99,

 

Please go with TAC's recommendation, because they have more detailed information about your issue.  As distance increases, the chance that you will get the theoretical maximum decreases.  You should not expect to get the theoretical maxiumum at 400 meters.

 

ok. Noted. Thanks for the advice. In your opinion, what is a reasonable range for the throughput?

Wolfie99,

 

There are quite a few factors that go into figuring it out.  If you have time, the Outdoor Mimo Networks VRD here:  http://www.arubanetworks.com/wp-content/uploads/OMWN_VRD_2012-01-04-1.pdf goes into great detail about that.  I would just end up copying and pasting that information here.

 

Thanks Joseph,

 

I'll go through that document in details.

Wolfie,

 

Are you using two wireless clients on each side of the 175 bridge to do your thorughput tests or are you doing wired tests? If you are testing wireless client-to-client iperf, then your numbers are about right (your two wireless clients are half-duplex hitting nearly half of the theorietical max). You will/should see much higher thorughput when you wire in both clients (one on each side). You also should be testing large frames and multiple streams. What iperf strings are you using to do your test?

 

In short, in case you are doing wireless client, change to makeing both sides wired, 4-stream tests with larger packets sizes. Also UDP tests will be faster than TCP. 

 

@400m with those antennas, you should be able to get well over 100Mbps unless your 5Ghz air is not clean. The outdoor VRD has a rate versus range graph to give you a good idea.

Hi Howard,

 

Thanks for the reply. Yes.. I'm having one laptop each on both sides of the mesh portal and point. Each laptop is connected wired to the data port of the mid-span injector where the MPP and MP are connected.

 

So when I run iperf test between the 2 clients, the mesh link is my transport medium. It looks like this :

 

 

switch ----- iperf client (wired) --> ====================== <-- iperf server (wired) --- switch

                                                             (mesh link 300/300 Mbps)

 

 

The iperf parameters we used at both sides are as follows :

 

Iperf –c 1.2.3.4   -w 512k  -t 60   -P 4

Iperf –s  -w 512k

 

 

I can achieve an UL / DL speed of 300/300 Mbps for the mesh link and a RSSI value of 38dbm. Theoretically, I should be able to achieve a throughput close to 150Mbps right? But I'm only getting around 77Mbps throughput. Is this acceptable?

 

 

Thanks

 

Sorry, a previous post you had mentioned something about the client's wireless NICs.

 

So your RSSI is good, great even, and your iPerf looks fine to me. The next question would be if there is any other RF in the area using the same channel, adjacent APs on the same channel, etc? If you change your mesh link's channel (say 161- or 153+, etc), do you still see the same speeds and channel utilization?

 

Your power injectors, are they Gig rated at 802.3at, and the mid-span's links are for sure running at gig speeds? 

Hi Howard,

 

Thanks for the prompt response. Really appreciate it. I did not check the port speed of the mid-span injector. I'll need to verify it. You have a point there :)

 

As for the surrounding, I did checked using InSSIDer and there're no other APs in the same channels. Anyway, the external antennas are mounted high on the rooftop so co-channel interference is pretty minimal.

 

Due to country restriction, we are able to use only channel 149+ and 157+. So the 2 pairs of mesh portal-point are using bonded-channels which are far apart.

 

Thanks

 

Just as an FYI, if there is another mesh link/bridge setup, InSSIDer would not show the bridge link, just something to be aware of in case there is another 5Ghz bridge setup. Your antennas are 60deg wide, which means they can pick up a large swath of neighboring RF.

 

You have two mesh bridges setup, are they both IAP-175s and do both bridges show the same uplink speeds across the mesh? You may want to open a TAC case and get them all the tech information for them to review. Usually it's an issue with interference when both sides are showing good RSSI and bandwidth. But TAC can work out the details and/or steer you to other possibilities.

 

Good luck, let us know what you find or what TAC steers your way. 

Hi Howard,

Yes. They are both AP175 and both mesh links are showing a good 270/300 link speed across. I'll open a TAC case on this.

 Will keep you posted on the outcome.

 

 

Thanks for the advice.