thank you, so the 25th users can't connect to the AP?

No, its not like that. Being a shared medium the 25th user will also get connected to the AP but th eoverall performance will start degrading as more users join the AP.

I am not sure but I guess there is configuration parameter to restrict the number of users who can get associated with the AP.

-> Sushil

I'm sure this isn't relevent, but I was reading release notes for 5.2 WLC and I think it mentioned somewhere about HREAP having a client limitation per AP. Something about now it is 100 but it used to be like 24?

The impression I got was that they actually did have a client limitation on certain configurations (but in 5.2 it has been increased) so that the system did not support that many users.

Hey Guys,

Interesting thread.

Is this "recommended" number for data? ie per b/g radio, or does it include a mixture of data and voice if you run Voice in the b/g freq.

Many thx

Ken

This is a recommended number per radio... but this is not set in stone. The number depends on traffic. If your users are just browsing the INTERNET or checking email, you can have more than that number per radio. If users are transferring big files it might be 2-5 per radio. Voice is another story.... This is from a doc:

The Cisco Access Point can support up to 27 active RTP streams for both 802.11g and 802.11a at a data rate of 24 Mbps or greater. To achieve this capacity, you must use U-APSD and have minimal WLAN background traffic and radio frequency (RF) utilization. The number of calls may vary depending on the data rate, initial channel utilization, and the environment. At 6 Mbps, the Cisco Access Point can support up to 13 active RTP streams. Using U-APSD instead of PS-POLL provides higher call capacity because U-APSD is more efficient and has limited management overhead. An RTP stream is bi-directional audio to and from a phone.

Hey Fella, How is ya :)

Digresing a little, but near the same subject matter.

So here are my tests.

Running APs in mixed mode, and I have performed over 180 individual tests in the last two days. Using IXIA tools.

I see an average thruput when loading up the APs of approx 12 M/Bits ps over the 180 tests on different APs. I transfer 800 Mbits to every AP.

Some reach 17-18 Mbps, and some (only a couple) reach 5 Mbps, the majority between 10 and 13 Mbps.

With these statistics and user avg thruput stats, I can make a desicion on how my user-to-ap ratio can be met?

How would that sound to you experts out there?

I know what I have tested in the last two days may not be the same tomorrow, but thats the nature of it right? I wanted to get a benchmark.

The one thing I am thinking about is turning off 802.11b across all 300 APs, but there is no gaurenttee that this will make things better right? It may be SNR ratios that bring the performance down, and not CTS/RTS phy controls right?

Many thx

Ken

Ken,

These test are okay to get a good baseline. There are many other test you can do but the main thing is how is user experience. When you do your testing, are younoting how many users are associated to that ap and if any 802.11b clients are on? I guess this can give you a good baseline as ling as the traffic doesn't change. For instance when you are testing, what are the users doing at that time. All it takes is a constant download by one or two users to knock down the bandwidth. But again your test should be good enough to give you an average of users per ap. YOu must decide the bandwidth you would like for each user. Test some locations with 802.11b turned off and get some readings and compare. The only issue with turning 802.11b off is do you need to support these users. If for example you have guest access, you would need to support these users.

Ken

What were the authentication and encryption mechanisms used?

WEP

WPA TKIP

WPA2 AES

How many stations were there associated with the AP?

What fragmentation level did you use?

Did the wireless cards use the power save feature?

Mark