7signal | The Wi-Fi Performance Company

A Data-Driven Comparison of Cisco and Aruba’s RF Automation – Part 1

by Veli-Pekka Ketonen
8 Comments

Wi-Fi RF automation is a hot topic. Not because it’s new (it’s not) but because there are strong believers and non-believers. The discussion lacks facts and measurement data, which we will cover in this 2-part blog.

7signal worked in cooperation with Avans University of Applied Science to characterize 802.11ac Wi-Fi performance and behavior in their environment. We compared and contrasted the RF automation behavior of two market leading Wi-Fi companies.

The Test Environment

The test environment consisted of two learning centers on the university’s campus.  The existing Wi-Fi access points were replaced with each vendor’s solutions for one week each. Wi-Fi performance was measured by 7signal’s system of Sapphire Eye sensors and Mobile Eye apps for devices.

Learning Center 0

  • Seven access points, of which 6 were measured
  • Aruba AP225, SW version 6.4.3.7 (1 week)
  • Cisco AP2700, SW version 8.1.131.0 (1 week)
  • One 7signal Sapphire Eye 2100

Learning Center 1

  • Seven access points, of which 5 were measured
  • Aruba AP225, SW version 6.4.3.7 (1 week)
  • Cisco AP3700, SW version 8.1.131.0 (1 week)
  • One 7signal Sapphire Eye 2100

Special focus was on 5 GHz channel selection operation.

Each week began with four fixed 20 MHz channels and fixed power levels, with no RF automation enabled. Channel power levels were equal in both cases, following the manufacturer’s recommended power level settings. All other access points in the Learning Center used this configuration over the testing period. Each test period lasted from the previous week Friday afternoon until following week Friday afternoon.

Tuesday evening, RF automation was turned on and all allowed channels (European channel allocations) were enabled. In addition, power control was also enabled. The default settings for RF automation were utilized and channel bandwidth was kept at 20 MHz.

Some interesting and important observations were made. We will label the vendors as A and B. You can have a little fun guessing who is who.

RF Automation Switched ON

Below are graphs representing the power levels from 7signal’s Sapphire Eye sensors. You can see on floor 0 and floor 1 how the levels quickly changed once RF automation was switched on.

Floor 0 Power Levels

RF automation switched on

Floor 1 Power Levels

RF automation floor 1

My Key Takeaways

  • RF automation substantially increased power levels in both cases.
  • Vendor A’s  initial static power levels appears to result in a higher received power level, despite similar settings.
  • After turning RF automation ON, Vendor A’s power levels remain higher than Vendor B’s.

A Closer Look at Channel Balance

One of the key promises for RF automation is to allocate channels in a balanced manner at any given time.  Let’s take a look at these allocations at different times.

The chart below displays the 4 static channels that were used before the test period, as well as during the test period, but outside our test environment.

channel-usage-before-the-test-period

Now let’s look at the average channel allocation with RF automation switched ON. In the chart below, both vendors are compared for each floor.

Floor 0 Channel Allocation

Channel allocation after RF automation switched on

Floor 1 Channel Allocation

Floor 1 RF automation channel allocation

My Key Takeaways

  • Vendor A balances the channels relatively well channels across the available spectrum. I see some room for improvement, however, as some individual channels still have a high number of BSSIDs.
  • Vendor B’s channel allocations are heavily concentrated to the lower channels. This is a very undesirable behavior.
  • After enabling RF automation, Vendor B still allocates channels 36-48 to APs in this area. This happens despite the fact that they are already used by other APs in the building. This leads to a high concentration at low channels.
  • Vendor A clearly performs better than Vendor B. Can you guess who is who yet?

Stay tuned for part-2 of this blog series where we will review the data for channel stability and air utilization and how both are affected by RF automation.

Tagged with:, , ,

Leave a Comment

Your email address will not be published. Required fields are marked *

8 Comments

rovinguser

Vendor automation has it benefits but given enough time and measurements IMHO a static channel and power plan will outperform any current vendor automation algorithm.

Reply

2 776 889 km2

Hey, that’s a clever way of thnnkiig about it.

Reply

rovinguser

addition to the blog post : Cisco fw used was : 8.1.131.0
version mentioned in article was in fact the Juniper/Trapeze OS version

Reply

Veli-Pekka Ketonen

Thanks Kees. Corrected to text.

Reply

Samuel Clements

I do hope that the specific Cisco RRM and Aruba ARM configuration options used will be posted. There is a myriad of flexibility in both environments so a ‘straight up’ Cisco vs Aruba comparison is going to be difficult to claim unless you include these specifics… As you’re well aware, defaults are meant as placeholders and rarely reflect best practices recommendations from either vendor.

Reply

Veli-Pekka Ketonen

Sam, thanks for reading the blog carefully and commenting.

Vast majority of the all organizations globally run default RF automation (channel, power) settings, which were also used here. I agree RF automation performance can be improved with tweaking, but only to certain point. For network engineer it is essential to understand how network operates with different settings.

I will publish more details of the settings as soon as I pull them together.

I may also write a third part to this blog, which compares some active testing performance data, like throughput.

Reply

Rocky

So which vendor is A ? Is that disclosed somewhere ?

Reply

2 776 889 km2

Rocky, thanks for your interest! Objective here was to show how remarkable differences there are between the two leading vendors. It matters a lot which Wi-Fi you run and how it’s configured. We are independent provider of Wi-Fi performance management solutions and have no interest to promote one Wi-Fi access vendor above other. Which one is which, we’ll keep that with us. Thanks for asking, though.

Reply

Get Social

More from Veli-Pekka Ketonen