Has anyone tested 802.11ac wireless yet?

[zag]

Our 802.11ac deployment is now complete with 40 x Aruba 225 transmitters all around the school.

 

I'll update when I get some feedback from users about the speeds and range of the new devices.

 

I see from the management portal that most users are still connecting under 450mbps but that's certainly better than the 65mbps maximum we had before

[DrCheese]

How much was it costing you per access point? We've got 40 + AP-93's at the moment and 15 or so AP-105's

 

The AP-93's only either 2.4ghz or 5ghz, the 105's run both. Rather annoying

[avenn]

Hi njc235 & craig4gon,

 

 

Some interesting points - did you know that the802.11ac standard is 5 GHz only. Not many of the sales guys tell you that the 2.4ghz bit is only 802.11n.

 

 

To me (looking at the macro level) it’s more of a steppingstone to 802.11ax which is looking at combining 2.4 and 5GHz and hopefully(please do!!) 60ghz but it isn`t standardised yet.

 

 

I also wonder if it will get canned like others have. Itmight be better if they merged, if possible/suitable the mobile (LTE etc.) and Wi-Fi(802.11...) standards - just a thought!

 

If you want a glimpse of the wireless future then have alook at this video on the BBC BBC News - How will the 5G network change the world?

 

 

If this doesn`t get your techie bits excited then I`m notsure what will!

regards

 

Aidan

 

 

 

 

 

[avenn]

Hi Zag,

 

 

Amazing connectivity speeds. Have you tried any throughput tests?

 

 

Not to sure whats available on macs. I use nuttcp on linux and win.

 

 

Apparently lperf works on mac osx.

 

 

It would be interesting to see.

 

 

regards

 

 

Aidan

[zag]

How much was it costing you per access point? We've got 40 + AP-93's at the moment and 15 or so AP-105's

 

The AP-93's only either 2.4ghz or 5ghz, the 105's run both. Rather annoying

 

AP-225's were £582.75 each

[Paid_Peanuts]

Like Zag we too have rolled out ac over the summer. We chose Ruckus for our implementation and installed 50 of the R700 access points.

We have these connected to HP 2920 PoE+ switches - even though they don't require PoE+ (we future proofed there).

 

We took out old Cisco 802.11g kit - the difference is night and day.

So far with the Ruckus kit I have seen up-to 800 clients connected at any one time - we allow students to connect to a BYOD network. There is no noticeable affect on the wireless for our clients when we do this. We do prioritise the traffic between the SSIDs within the Ruckus controller, we have turned on the client isolation and their firewall - which restricts where the BYOD devices can reach.

Edited December 8, 2014 by Paid_Peanuts

[Jose]

I have tested a cisco 11ac ap with our managed cisco wireless.

Using dell 3340 laptops

 

With one laptop connected and copying over a folder with large iso files, the average speed was 30MB

So it seems just like with the old wireless standards, you need to divide by 3 to get the real speed.

 

When we had 24 laptops all connected and copying the same folder, they copied over at around 5mb

so we lost some speed to overheads on the ap.

 

Still a big improvement.

[njc235]

Seems very slow to me. What switching are you using?

[Jose]

switching is fine

 

Its the correct speed that I would expect.

 

Its important to do real world tests with wireless points, this means testing the speed by copying large files and letting the speed settle.

 

I always divide by 3 , 11g is real world speed is around 18-19, only if you do real tests do you find this out. I am quite happy with 11ac tests just do not believe marketing hype with wireless products.

[zag]

Yep we've had wireless 802.11ac for a year now and don't see huge speeds in the real world. But it is a big difference from the old 802.11n version.

 

About twice as quick from experience.

[psydii]

What I really want to see is a 3D deployment of 6+ AP bake off against a selection of vendors include the single channel one, with with 90 adjacent laptops doing running a test that looks like continuous loading/syncing of "roaming profiles" (which imho is a make or break for wifi load testing), and 30 ipads doing a web browsing load, and 6 ipads streaming to apple tv. With a timed spikes where 1,3,9,27 clients across the fleet try to print a graphical image, or save a large file.

 

Some devices should be running voice calls (SIP rather than something proprietary like Skype), and a couple video conferencing.

 

Vendors would be allow to specify the wireless configuration to show off their kit the best, but firmware would need to be posted to their website, and nothing involved in the final test configuration would be allowed to be covered by an NDA.

 

Anyway we can but dream. Best I've seen is the 1 AP bake done by wlanpro.com, but a 1 ap test is very far from an decent approximation of the real world.

[jtotheb]

What I really want to see is a 3D deployment of 6+ AP bake off against a selection of vendors include the single channel one, with with 90 adjacent laptops doing running a test that looks like continuous loading/syncing of "roaming profiles" (which imho is a make or break for wifi load testing), and 30 ipads doing a web browsing load, and 6 ipads streaming to apple tv. With a timed spikes where 1,3,9,27 clients across the fleet try to print a graphical image, or save a large file.

 

When we were looking at upgrading our setup we did a 5 AP test over 2 floors with the vendors we were seriously looking at. (The time it took to set up and conduct meant that we couldn't do more than a few) The final stress test had 85+ devices constantly streaming while 2 class sets of laptops (20+ in each) simultaneously logged on. We tried to make it an extreme version of a "real world" scenario.

[DrCheese]

Yeah - I'm not sure any vendor would agree to that as if they came near the bottom it would forever dent their sales

[psydii]

@jtotheb nice setup. 11ac? Which vendors did you look at and which one did you choose? How did you monitor/measure performance?

[jtotheb]

We did test some ac products but we ended up going with Ruckus and their top dual-band n points as we don't, and won't until this system is due for a refresh, have enough ac clients to take any real advantage.

 

The main measuring tool was the good old stopwatch for the how long will this take question. Test, re-test and verify which took a fair while. We went through plain speedtests and then moved onto what we considered "real" scenarios e.g. Classroom A had everyone streaming iPlayer live whilst Classroom B all logged on, that sort of thing. The final test involved 135 devices in more or less class sets (between 20 and 30 per room) all connected and all hoofing traffic through simultaneously. The tests were artificially hard in some sense as you'd expect to get some stagger in logons IRL.

 

It may sound like a daft thing to say, but it wasn't until we were there with a stopwatch that I truly appreciated how much better the 5ghz band performed.

[DrCheese]

^ Yeah, 5ghz is leagues ahead of 2.4. Annoyingly most of my points can only do either 2.4 or 5, not both at the same time so I'm stuck on 2.4 only mode until we replace our wireless network or 2.4 only devices disappear