Ruckus Test Results - Is this right?

[eean]

We have 2 Ruckus APs + 1 AP controller on trial. One is a/g/n the other is g/n. We're using Macbooks which are all a/g/n. Encryption: WPA-PSK.

I was very excited about Ruckus, and after by test results I'm feeling less excited.

Question 1. What is the SpeedFlex test actually reporting?

I've been comparing results from SpeedFlex with iPerf. iPerf is a simple program - it runs on the server and on the client then fires a load of data between the two and times how long it takes (it doesn't write data to the HD).

I realise that it not an exact comparison: iPerf is testing the whole route between the client and the server whereas SpeedFlex is just testing the speed between the AP and the Client.

 

Speed between AP and Controller - Speedflex: 91.3mbps iPerf: 74.9mbps.

So the uplink to the server is not brilliant, but not the limiting factor when you look at the results.

 

With an a/g/n access point and the client (on a/n) in the classroom - SpeedFlex: 80, iPerf: 31.2

With an g/n access point and the client (on g/n) next door - Speedflex: 56.2 iperf: 23

 

With an a/g/n access point and the client (on g/n) next door - Speedflex: 56.2 iperf: 23

 

With an a/g/n access point and the client (on g/n) next door - 8 other users - Speedflex 48mbps iperf: 1.8-3.41 (OK there was obviously some data transfer going on by the other clients - but it was only web and we've only got a 4meg line!)

 

I know that there is overhead with the encryption, but I was expecting better 'real life' throughput from n than that. The best I got was 31.2. I can get better out of a consumer grade Belkin (35). People on the web say you can get 80mbps real life speed out of n.

 

Question 2. Shouldn't proper n use both the 5ghz and the 2.4ghz frequency if it can?

The dual band access point balances the clients between a/n and g/n implying that it's one or the other, not both. I thought n used both simultaneously?

 

Question 3. Shouldn't the AP controller sort out which AP is on which channel?

We had it set to auto but it put two APs which were in neighbouring rooms on channel 1. I had to move it to channel 6 manually.

 

Question 4. How long does it take to move the clients over to a better AP.

It had clients balanced between a AP downstairs and one next door. We unplugged the one downstairs and the clients all switched to the one next door - they didn't even notice!

However, when we put the AP in the same room it took a good 5 minutes (after it'd booted up) before even one client connected to it.

 

Question 5. Why is there an 'unknown' SSID attached to one AP?

We only have one SSID set up. However, using inSSIDer I noticed that one AP had two SSIDs -ours and another called 'unknown'. It was definitely Ruckus and it reported the same or slightly lower signal on the same channel.

 

Any Ruckus experts or real life users able to shed light on any of these would be much appreciated. I want to be sure that this is the best system that we can afford rather than one that we buy on the basis that other people say 'yeh, it's good'. TBH I wasn't wowed!

Edited September 18, 2009 by eean

[Tricky_Dicky]

Q1)Haven't used it to be honest, never saw the point in putting more strain on bandwidth. I took the opinion that if the clients find it working quickly then it is quick enough.

 

Q2)I think "a" uses the 5GHz spectrum, "n" is still in 2.4GHz.

 

Q3)It will sort it out for you, but it won't do it immediately. It appears to scan every 10 minutes or so and will adjust not only the channel it transmits on but also the amount of power it is transmitting, reducing the power or changing the channel as it sees fit. This works very well indeed.

 

Q4)Moving clients on to a better AP is a local configuration on the client. It depends on how high the roaming is set to. The Intel chip sets have the values available to change in the advanced properties.

 

Q5)All I can think of for this is that you have set up 2 SSIDs or the guest access it turned on but not configured. If you are using the newest software (8.1) then it could also be the hotspot software.

 

I installed the system at our large Primary school over the summer and after a disastrous first lesson with the kids updated the software and it is now working so well it's frightening. It basically manages it's self and no one even notices it's there it just works which is exactly what I wanted.

Good luck

 

Rich

[eean]

Q1) I took the opinion that if the clients find it working quickly then it is quick enough.

I fear it's not working quick enough. I opened a few pictures into a paint editor - it was almost instant on a wired connection but took over 45 seconds on wireless. Now imagine 24 kids doing that at once.

For web it was fine - but so was our old gear.

 

Q2)I think "a" uses the 5GHz spectrum, "n" is still in 2.4GHz.

n can be either (if you have a dual band AP). However, I thought it could bundle both simultaneously, which it can't.

 

I also thought it used wider bands (i.e. 2 channels) to achieve it's top speed (which is why it works better on the 5ghz spectrum). I didn't spot it doing this on my frequency monitor - any one know if it does, or why it didn't?

EDIT: I've just spotted the 'channelization' option is set to Auto. Should that be 40? Why wouldn't it select 40 itself?

 

Q3)It will sort it out for you, but it won't do it immediately.

That'll be it. Thanks.

 

Q4)Moving clients on to a better AP is a local configuration on the client. It depends on how high the roaming is set to. The Intel chip sets have the values available to change in the advanced properties.

I thought that was the point of tha Access Point controller - that it chose the AP rather than the client, which is how the load balancing works and which is why a managed solution is better than an unmanaged one.

 

Q5)All I can think of for this is that you have set up 2 SSIDs or the guest access it turned on but not configured. If you are using the newest software (8.1) then it could also be the hotspot software.

No, that's not it. We are using the older 8.0.1 firmware, so that might have something to do with it.

 

 

Thanks for your help.

Any more help on these issues would be appreciated.

 

Thanks.

Edited September 23, 2009 by eean

[Tricky_Dicky]

I'm not sure what's happening there then. But it depends what you are comparing it the wired and wireless speeds too. If you have a Gb connection plugged into a laptop and then change over to a G connection then it will be slower and no wireless solution in the world is going to be any better.

 

Something to think about though is what exactly are the laptops going to be used for? When I first started I got given lists upon lists of programs which "must" be installed on everything because they are used "all" the time. Having installed almost none of them, no one has mentioned them as being missing! These days almost all teaching that I witness using laptops is based on the internet using web 2.0 applications or MS Office. Maybe it's different in your school but that is my opinion.

 

I stand corrected on the frequency, it can be either, but not both. I have no idea why it wouldn't select the wider band, try logging a call directly with Ruckus and see what they say.

 

A managed solution will do that to a point but only when it hits a certain level. Lets say the access point and client are still communicating at 10Mb/s, if the client is set to not change unless it has to then it wouldn't bother.

 

Best of luck with which ever solution you decide to go with and keep me informed of how you get on!

 

Rich

[White_Fi]

Hi eean,

 

The speeds between the client on the dual band a/n are not high enough at all! I have never received anything lower than 175Mbps

Have you enabled Channelization to 40MHz, this will double our speeds and can be enabled on the 2.4GHz when the AP can speak 802.11n and the Dual Bands again 802.11n and 802.11a.

 

We can also specify whether the AP is broadcasting on the 2.4GHz via using WLAN groups.

 

Two AP's in neighbouring rooms on the same channel should make little impact on each other as the interference is avoiding each AP, the fact they both chose to use channel 1 is more than likely down to rouge analogue noise on the other channels so the AP's choose channel 1 as the quietest to sit on and then steer the RF energy away from each other.

 

Think about client roaming aggression if this is set to low or medium the client will wait for the AP to come up before deciding to move to a new AP, try upping the aggression level to a higher setting and test again.

 

As for the unknown SSID, do you have MESH enabled on the AP's? if so this "Unknown" SSID is the SSID used for the AP's to MESH together.

 

Kind Regards

Stuart White

Net-Ctrl

[eean]

The speeds between the client on the dual band a/n are not high enough at all! I have never received anything lower than 175Mbps

Have you enabled Channelization to 40MHz, this will double our speeds and can be enabled on the 2.4GHz when the AP can speak 802.11n and the Dual Bands again 802.11n and 802.11a.

175 was the kinda speed I was hoping for! (I'm guessing that's SpeedFlex speed - is that with or without encryption?).

 

I tried enabling 40mhz on Radio A/N (5ghz). On 2.4ghz it only has 20 as an option.

After doing so, my clients would only connect to 2.4. When I disabled 2.4 on the client they would not connect at all. This could suggest a client issue, however Macbooks can reportedly support 40 and when running Windows there is an option to disable it.

 

Two AP's in neighbouring rooms on the same channel should make little impact on each other as the interference is avoiding each AP, the fact they both chose to use channel 1 is more than likely down to rouge analogue noise on the other channels so the AP's choose channel 1 as the quietest to sit on and then steer the RF energy away from each other.

 

Even if they are both serving clients in the same room? Surely they can't avoid interference if they are each is connected to clients virtually sitting next to one another?

 

Think about client roaming aggression if this is set to low or medium the client will wait for the AP to come up before deciding to move to a new AP, try upping the aggression level to a higher setting and test again.

But I thought the client didn't really control which AP it connects to but it was the AP controller that decides.

In this thread you say:

It would load balance the clients as to how busy they are / how many clients happen to be on one AP at any one point. If a client is wanting to stream video or make a VOIP call on an AP that has 25 other busy clients then the client wanting to VOIP would roam to a less busy AP.

Would the client decide it roams or the AP controller makes the client roam?

I'm not trying to be awkward - just if it's the client then how would it know, surely it's almost accidental. If its the AP controller that decides then that's clever!

 

As for the unknown SSID, do you have MESH enabled on the AP's? if so this "Unknown" SSID is the SSID used for the AP's to MESH together.

That'll be it.

 

Your help is really appreciated. Thanks. I have high expectations of how intelligent the system can be and I need to be 100% certain it's right before spending a big pile of cash on it!