Wireless Access

Hello!

I'm at testing 6.3.0.0 vs. 6.1.3.9, using two (reference) laptops.
APs: 125 and 135; Controller 3600
DELL Precision M6300 (Windows XP SP3)  DELL Wireless 1505 (BCM2328/BCM4205500)
DELL Precision M4600 (Windows 7 SP1)  Intel Centrino Ultimate-N 6300 AGN
5Ghz, 40MHz channels, single client, absolute clean environment - no other interference or clients.

With both clients and both sorts of APs the maximal download speed is only half with AOS 6.3.0.0
compared to AOS 6.1.3.9 with otherwise same (extreme clean) environment!

HELP!

Any experiences out there?

#3600
#AP135

Not running that, but will be watching to see what you find.

 

Seemed like we had some coverage "shrinkage" when we moved from 5.0.3.3 to 6.1.3.6.  Complaints would go away when I rolled back to 5.0.3.3 or moved forward to 6.1.3.8, so problem seemed isolated to 6.1.3.6, so very interested to hear of any problems with certain releases.

You've got a clean and controlled test bed. I'd log that with Aruba TAC if I were you?

>You've got a clean and controlled test bed. I'd log that with Aruba TAC if I were you? I have already made this yesterday afternoon... As far as I can see at the moment only peak performance of a single user / single tcp (http) downstream per AP is affected. As you start a second tcp (http) downstream the summ of the performance is as before (up to 6.1.3.9). (MTU,MSS,GRE-tunnel,CPSEC,ip-fragmentation?) I am curious what Aruba will say...

There is of course yet no answer of Aruba. But, the APs 125 and 135 do not seem to be strong enough for 6.3.0.0. With one single client 100% of CPU!!!

 

High utlization on the AP?

Yes.

 

Can you please share the show ap debug system-status ap-name <apname>

AP125 with one active client.

 

Currently running 6.3.0.0 at 4 of my sites but I did not see any obvious changes in coverage or performance. CPU and memory on my busiest APs with 10 plus clients is 1-5% cpu and 30% or lower memory.  Sites use AP105s and 6000, 3400 and 3200 controllers.

 

 

Hi,

 

Also "client match" does not function properly.

No miracle with 100% of CPU in AP.

e.g., the neighbour list of an APs is suddenly empty... and so on...

 

Will AP12x (and partial AP13x) really be able to be supported with version 6.3?

 

I will change back to 6.1.3.9 and wait what Aruba will say...

Hi,

 

>and wait what Aruba will say...

 

Aruba has confirmed the problem and is working on a solution.

 

Regards

 

any updates to this from Aruba or otherwise?

I am running AOS 6.3.0.1 on a 7200 controller with AP-135s. I cannot downgrade to AOS 6.1.3.x because I need Apple Bonjour services on our WLAN.

 

Yesterday I migrated a building on our campus from Cisco 1142 APs to the 135s and today I get complaints from users with Centrino Ultimate-N 6300 and Centrino Advanced-N 6205 wifi cards. They are using Debian Linux as their operating system.

 

 

So I'd be interested in a solution as well!

 

Is a fix for this behaviour something that is on the roadmap for the next 6.3 AOS release?

I have tested, in the meantime, also AP225 at 3600-Controller with version 6.3.1.1 and with 802.11ac 3SS (1300 MBit/s). With also catastrophic download performance on SSIDs with encrypting (WPAx OR WPAx-PSK). Upload and not encrypted transference (open) are nearly not affected! The maximum download-transference rate amounts to about 20 MByte/s. - I think for ALL USERS OF A 3600-controller TOGETHER! Aruba-TAC is informed and checks this at the moment. Regards, Hardy

Hi! News: A Patch (from TAC) to 6.3.1.1 (will be included in the upcoming 6.3.1.2) has solved the 802.11n performance issue (Intel N6300 Centrino Clients) with (AP125),AP135,AP225. We have completely moved to AP224/224. 802.11ac performance is still bad and unter investigation.

UPDATE:

Now we are running: 3600-Controller, AOS 6.3.1.2, AP224/225

After the update to 6.3.1.2 almost all problems are solved.
Now the data throughput is in the scale as expected..
However, many client drivers still have problems and are probably improved in the next time.

With 6.3, the AP radio is going to scan more aggressively compared to 6.1.3.9, there's a significant difference in terms of

the channel scan rate. If clients are PS enabled, it is only going to make things worse. Could you tell us more about what exactly is the measurement factor? It'd enable us to respond precisely.

 

i.e., TCP iperf/jperf test run from one WLAN client to another, where both clients are associated to same BSS
or TCP based download from wired source?

 

Also, are the 6.1.3.9 and 6.3 codes running across two different platforms which are benchmarked side by side (with default config, of course) or are they iterations on the same test bed?