Technology Blog

The 802.11ac Grand Prix. What’s Under Your Hood?

by Moderator on ‎11-20-2013 12:35 AM - last edited on ‎11-20-2013 09:59 AM by Chief Airhead Chief Airhead

Alan Kay once said, “People who are really serious about software should make their own hardware.”  Steve Jobs quoted Alan when he first introduced the iPhone and in recent discussions with Aruba’s engineering team, this same philosophy applied to the development of the 802.11ac AP-225.

 

To achieve the design objective of making the AP-225 the highest performing 802.11ac AP, the hardware and software development efforts were united. This allowed for new and innovative capabilities like having multiple uplink interfaces on the AP with software that aggregates the links and the ability to use dual-core processors with software for symmetric multiprocessing. It’s these types of innovations that make the AP-225 the king of the road.

 

Unlike Aruba, Cisco took a different path to 802.11ac. Kudos to them for recognizing the need for 802.11ac APs as clients are overwhelming today’s Wi-Fi networks. It’s probably why they announced two 802.11ac APs within a couple of months of each other.

 

When you look under the hood of the AP-3700, it’s using the exact same Freescale 1023 Single-Core 800MHz processer as its predecessor, the 802.11n AP-3600. It’s like painting racing stripes on a Yugo and calling it track ready. CPUs are the AP's engine and faster CPUs mean more clients per AP.

 

 

YugoAP.png

 

 

In an attempt to show the Cisco AP-3700 outperforms the Aruba AP-225, which repeatedly outperformed the AP-3600, Cisco paid Miercom to cook-up some tests with improbable scenarios.

 

Upon initial review of the Miercom report, we noticed they used custom engineering builds from Cisco, which are not publicly available. For Aruba, they didn’t even bother to download the latest publicly available software, which was accessible to them. I guess they had to put some bananas in the tailpipe for a fair test. 

 

Lets explore Miercom’s TCP throughput tests.  The new report shows TCP throughput for the AP-225 at 705Mbps for a Dell 3SS laptop. Interestingly enough, Miercom’s previous report showed Aruba’s TCP throughput at 768.83Mbps for the Dell 3SS laptop. I’m puzzled as well as to how Miercom came up with two different results for the same test. Download Miercom’s test report.   

 

 

Miercom Test Report: October 2013

1.png

 

Miercom Test Report: November 2013

2.png

 

 

Next up. Multi-client testing. Miercom must have put the AP-225 in Valet Mode as it only got 40Mbps for their 60-client test. Our own internal test results show 346Mbps in bi-directional throughput tests. But, don’t take our word for it. The University of Delaware reported in a Network World article that their AP-225s sustain 400Mbps on a single AP with over 120 clients.

 

Aruba Test Report: November 2013         

3.png

 

 

Miercom Test Report: November 2013     

4.png

 

 

So, when you’re about to kick the 802.11ac tires, don’t be fooled by the racing stripes. Take a good look under the hood and make sure it has the processing power you need for today’s high-density environments.  Learn more about Aruba 802.11ac with ClientMatch. 

 

 

Comments
Moderator
Nice write up

Every time I see Gerry in the office, I can only think one thing...

 

 

 cheatingcisco.jpg

banana.gif

Ryan Bonner

An interesting write-up. Are there any other data points that the University of Delaware test contradicts, in regards to the Cisco test?

Moderator

Hi Ryan. 

 

The interview with the University of Delaware was conducted by Network World. Aruba was not involved, so I don't have any additional data points to share other than what was published. 

 

Cheers, 

Gerry

 

 
I'm a fan of both Aruba and Cisco, but both marketing camps are in 11ac overdrive. We've seen the bashing from both sides. While you point out the Miercom report detailing the Aruba 225 vs Cisco 3600 11ac module, Mirecom clearly shows the 225 out performing the module. Here Mericom clearly shows the 225 is out performed by the 3700 so says Miercom. 
 
 Is Miercom a "buy" my services and you win ? Or did Miercom get it wrong ? 
 
The take away here while marketing is marketing. Let's not forget you guys ALL live in glass houses.
 
 
Would you like an honest test. I've been testing the 3700 and know it every well. Send me a demo setup and Ill be glad to do a drag race .. Let me look under the hood .. 
 
 

"

 

Upon initial review of the Miercom report, we noticed they used custom engineering builds from Cisco, which are not publicly available. For Aruba, they didn’t even bother to download the latest publicly available software, which was accessible to them. I guess they had to put some bananas in the tailpipe for a fair test. 

 

 

"

 

The cisco 3700 isn't released to the public, so there is no code available to the public. Not sure what you mean by custom code. 

Guru Elite Guru Elite

Gstefanick,

 

Peer review is normally done with publicly obtained hardware and software, so that others can replicate it and have a level of comfort.  Private code also opens the possibility of having code that is only optimized for the test being performed but otherwise does not function.    If a test is performed, but the uninformed customer cannot obtain the hardware to validate the test, they continue to be uninformed.

While I agree with your response this article is misleading. The 3700 isn't out yet and this is why the report referenced (beta code). Just adds confusion. 

Guru Elite Guru Elite

Gstefanik,

 

 

If a product is unreleased, it is probably best that they just say the numbers they got for their own product and do not compare to anyone else.  That way their customers know they are on track, and that it even could get better.  That is not what happened here and they exposed themselves to scrutiny by attempting to compare it to a product that everyone has access to.  The result was out of line with Network World's interview of an actual customer, and previous performance numbers.  That would have a reasonable person crying foul.  That is what I got out of the article.

Aruba Aruba

Here's another report you may have seen, though it is not testing 802.11ac.

 

The University partnered with Ruckus so the results are biased towards them, but in most cases the Aruba came out pretty well.  I'm a bit dubious about the channel selection test, but then again I didn't configure the Aruba for them.

 

link

Guru Elite Guru Elite

Michael Clarke,

 

The problem, in general with that report resides in the document -

 

"Ruckus Wireless, of Sunnyvale, CA, is one of several WLAN vendors with whom the School of
Information Studies has a close working relationship. Ruckus provided the equipment for the tests in this
paper and also funded a grant at the iSchool to improve the state of Wi-Fi education."

MVP MVP

I balieve this test should be done on an event like a wireless tech field day and in which each company bring their best  engineer  configure their own gear... This way there wont be any misconfiguration or any excuse about the results..

 

Anyways i dont know if wireless vendors would agrree on this kind of test...

 

I dont like 3rd party test... they can be bough... neither someone from cisco doing a test of aruba vs cisco or someone from aruba doing a test  of cisco vs aruba...

 

Just my 2 cents.

 

Cheers

Carlos

Search Airheads
Showing results for 
Search instead for 
Did you mean: 
Announcements
Read all about it! If it’s happening now, it’s in the community.

Check out the latest blogs from your community team, the community experts and other industry sources.
Labels