“Horses 2” by Steve – originally posted to Flickr as Horses 2. Licensed under CC BY 2.0 via Commons.
I learned recently that horses having eyes on the sides of their heads means they were historically hunted in nature. Like rabbits, they rely on peripheral vision to detect dangers.
When a horse is racing or pulling a carriage, they can get easily get distracted or startled. So this is why they are outfitted with blinders (sometimes called blinkers). To focus their vision to the front, so they don’t get distracted. This reminded me of a problem we tend to have in the WiFi world.
Yes, I’m saying AP’s are kinda like horses. -Jason Grant
Don’t believe me? Consider the following example:
Here we have a decent sized floor with WiFi designed to support a large number of users and devices. Just for this example, we’ll use 2.4 GHz, where we only have 3 non-overlapping channels: 1, 6, & 11. To make it easy to see, I’ve colored the AP’s with the same channel.
See that part I circled in red? This is an area that is covered by multiple AP’s that are using the same channel. in a blog article described it well.
Remember, an AP is like a horse. It has great peripheral vision. It can hear client devices, on the same channel it’s on, that are actually connected to a different AP! See the trouble? An AP needs to read in all client transmissions on it’s channel, even if it’s a transmission intended for a different AP.
Eeek! What do we do? In comes another Cisco Wireless feature called Rx-SOP. This stands for Receiver Start of Packet. Since the marketing folks haven’t come up with a fancier name, I’ve come up with a cool-looking logo for it. What do you think?
How is Rx-SOP like putting blinders on a horse? Great question. Just like blinders restrict a horses vision to just what the operator thinks is important (straight ahead), Rx-SOP allows an AP to ignore incoming packets that aren’t strong enough. This effectively allows you to shrink your AP cell size to minimize co-channel interference.
How to Turn Rx-SOP On
First thing to remember is that Rx-SOP shrinks your effective coverage area. If you shrink it too much you’ll make coverage holes.
|802.11b (2.4 GHz)||802.11a (5 GHz)|
|Auto||Radio Default||Radio Default|
|Low||-85 dBm||-80 dBm|
|Medium||-82 dBm||-78 dBm|
|High||-79 dBm||-76 dBm|
To configure the RX-SOP threshold go to Wireless >Advanced > Rx Sop Threshold. By default Rx-SOP is on “auto” which means it’s effectively off. Each radio is set to Auto, Low, Medium, or High.
The level you set hear instructs the AP to literally ignore incoming packets that are quieter than the configured level.
If you want to use Rx-SOP for only a few AP’s you can use RF profiles. How to do that is described in the High Density Experience (HDX) Design Guide.
HDX is acronym speak for Cisco’s High Density Experience, which is marketing speak for a series of patented features that specifically address the challenges that stem from the ever growing number of WiFi clients and their need for more reliable, higher bandwidth network access.
The truth is in order to be WiFi Compliant, especially in the early days, certain things needed to happen by default. These default settings may have worked a few years ago but today, even though you’re using high-end AP’s and trusted code levels, you may be getting unacceptably poor performance. This is why I wrote up 3 Steps to Tuning a Cisco WLAN Controller From Default Settings. In just a few minutes of configuration time and a radio reset, you may experience exponential performance gains.
Tuning a WLAN controller only solves part of the problem. This is especially true as your deployment begins to take on more clients, demanding more bandwidth. What kind of symptoms will you experience?
- Clients are still in charge of deciding when and where to roam.
- If you elect to use them, 802.11ac introduces larger channel widths meaning we’re more vulnerable to interference.
- With WiFi only one client can talk at a time… it’s half duplex. As clients get further from an AP they begin to talk slower to stay connected, leaving less time for fast clients to talk fast.
- AP’s are getting closer together to accommodate more clients, and there are limited RF channels we can use. Even with the newer FCC rule changes adding more available channels, it’s mostly good news for 11ac environments.
To solve these problems we need Optimized Roaming, CleanAir for 80 MHz Channels, ClientLink 3.0, Air Time Fairness, and Turbo Performance! Let’s dive in to what buttons to press, which options to check, and how to get it all working for us.
1. Optimized Roaming
What’s the problem: Client stickiness. The 802.11 standard says clients can roam. As such, it’s the clients responsibility to decide when and where to roam… they just don’t do it very well. Normally the client will not even TRY to roam until it can no longer connect at the lowest mandatory data rate. This is why I recommend turning off the lower data rates, suited to your environment. We are preventing the client from being really bad, not promoting good behavior.
What it is: Optimized Roaming promotes good behavior. If the client’s data signal strength dips 6 dB the AP will send a disconnect message, prompting a roam. You can optionally set a minimum data rate. If the clients tries to dip below the minimum, it’ll get a disconnect message, prompting another roam. Use caution here. If the client cannot connect at the minimum data rate, it won’t be able to connect at all.
How to turn it on:
- Wireless > Advanced > Optimized Roaming
- Check the Enable box in the 802.11a and 802.11b sections
- Note you’ll get a warning: Modifying the default settings for the Optimized Roaming data rate and CHDM RSSI configurations could result in unintended client connectivity problems. Please be careful when making changes from the default settings
- Reference from the 8.1 Configuration Guide
2. CleanAir for 80 MHz Channels
What’s the problem: It’s great that 802.11ac provides for greater bandwidth. It does that by using more spectrum, the channel is wider. More channel space means vulnerability to more interference. Interference is that that invisible threat that decreases overall throughput which often presents during times of load and is very difficult to troubleshoot.
What is is: CleanAir for 80 MHz Channels uses the built-in spectrum analysis of the Cisco AP to discover, identify, classify, and, most importantly, mitigate interference. Right away.
How to turn it on: CleanAir is OFF by default. So step one is Turn It On. It will automatically detect, identify, and classify. What we need to turn on is turn on 80 MHz channels and the Mitigate part of CleanAir. To do that got to Wireless > 802.11a/n/ac > RRM > DCA. There change the channel width to Best and at the bottom you need to make sure Event Driven RRM (ED-RRM) is Enabled. Note: By selecting the “Best” channel width, you are allowing the system to automatically back down the channel width if interference does not allow high-performance at 80 MHz.
3. ClientLink 3.0
What’s the problem: As a client gets further from the AP the signal strength from that AP get weaker. As a result, because it’s not yet time to “roam” it ends up lowering its data rate. It talks slower in order to stay connected. WiFi is a shared medium, meaning only one station (client or AP) can talk at a time. If a client is talking slow, there’s less time for the fast talkers to talk fast.
What it is: It’s fun with physics and it doesn’t need any client-side intelligence to work. So 802.11n introduced something called MIMO. It’s basically some fancy antenna technology that allows the AP to communicate with a client using multiple antennas at once. As seen in this 1 minute video called ClientLink at the Beach, when the transmissions from two or more antennas meet, they create a sweet spot of coverage. If your client is in that sweet spot, they’ll experience higher signal strength, and likely connect at a higher data rate. What is DOES is detect when the client is not in that sweet spot and it starts to manipulate the transmission timing on the multiple antennas putting a sweet spot right on the client.
How to turn it on: It’s on by default and works with any WiFi client out there that supports 802.11 a/g/n/ac!
4. Air Time Fairness (ATF)
What’s the problem: Guest users can potentially take up their unfair share of bandwidth downloading app updates or movies, degrading your business related WiFi traffic. Per-user bandwidth contracts only go so far. Because clients connect at different speeds, simple rate limiting is not efficient for WiFi.
What it is: Cisco Air Time Fairness allows us allocate an ability to consume bandwidth based on a defined group such as guests or smart phones. Regardless of a clients connected data rate, it can only talk a certain percentage of the time.
How to turn it on: Since you need to identify your groups and allocations, it’s not on by default. It’s a good idea to run ATF in monitor mode to understand actual bandwidth usage by SSID, by AP, or by a group of AP’s. Understanding that you can define and apply a policy.
- Turn on ATF Monitor Mode
- Configure ATF Policy
Here’s where a bit of planning helps. Remember that a policy is applied to an SSID, to an AP, or to an AP Group. First take a look at your actual Air Time allocation. Click on Wireless > ATF > ATF Statistics. Select an AP you want to see Air Time stats on. The top table shows you allocation over the last 3 minutes. The bottom table shows allocation as long as the radio has been up. Use this to determine the relative impact by enforcing “fairness.”
In the Wireless > ATF > Policy Configuration menu you will notice a policy ID 0 called “Default” with a weight of 10. This default cannot be changed, but don’t worry, you don’t need it to. Define all of the policy elements here. So if you want 75% employee allocation and 25% guest, you’d create 2 elements with appropriate weights. If you are building policies for multiple SSID’s or AP’s, just group them all together.
- Apply the policy
Click on Wireless > ATF > Enforcement SSID Configuration. If you’ve followed the steps above, your radios will be in ATF Monitoring mode. Here is where you with them to “Optimized” mode or “Strict” mode enforcement. As it sounds, Strict mode enforces hard limits. Optimized allows for spiking above the policy if possible.
In the Policy Enforcement section you specify an SSID and a ATF Policy ID to enforce.
For a way more technical description check out the Cisco Air Time Fairness White Paper.
5. Turbo Performance
What’s the problem: As your network scales to many more devices with greater bandwidth needs, it won’t matter how fast we’re connected if we can’t process packets. With 802.11ac we’re processing nearly double the packets as with 802.11n! We could address this with a faster CPU but that’s an expensive upgrade.
What it is: Commodity hardware and chipsets need to have a central CPU to process packets. With Turbo Performance, Cisco dedicates CPU & RAM to each RADIO to perform packet processing at the edge. This is much more efficient than having a central processor do everything.
How to turn it on: You’re in luck! You have nothing to do to take advantage of this feature.
So this HDX… High Density Experience… is more than just marketing speak. It’s a set of patented technologies that show how hardware is just as important as software and something only Cisco can deliver.
- Optimized Roaming
- CleanAir for 80 MHz Channels
- ClientLink 3.0
- Air Time Fairness
- Turbo Performance
For more information check out http://www.cisco.com/go/80211ac.
It really feels to me that here in the US there’s an abundance of tech jobs and they’re being filled with fresh talented people that don’t think the same way I did when I entered the workforce so many years ago. And that’s GREAT! When I got my start, I dove in to protocol analysis and deployment, and network design. In my first professional network engineering job, I had a great mentor that taught me network operations. It was extremely manual, extremely organized, and worth 100% of the effort.
In my job, I help and interact with smaller businesses, typically fewer than 1,000 employees. Considering what they’re asked to do, these companies have some of the brightest IT minds I’ve come across. I need to remember they come from a different place.
Here’s what I mean: I still have music I ripped from CD’s and I constantly play the game of which external hard drive it’s all on. Today thanks to Apple and others, my playlist is nothing more than a list of songs I have rights to listen to and it really doesn’t matter if it’s played from my device or from the “cloud.” I cherish these experiences and I learn from them, not to keep them alive, but rather to pass on these lessons.
The same sort of thing is true in the networking world. Cloud-managed networking systems are very powerful and optimized out of the box. It’s a solution that meets some needs many people weren’t aware were even there. Still there’s a huge need to proliferate and maintain on-premises network deployments, and that’s where I often find pain. This pain has a name. Its name is “Default Settings.” Case in point: Cisco offers arguably the best on- premises wireless LAN products in the industry; yet day after day I talk to smart network engineers feeling genuine pain. This is precisely why I authored 3 Steps to Tuning a Cisco WLAN Controller From Default Settings. Default settings are there for a reason and often times that reason is not to maintain optimum performance. In the case of 802.11, the default settings are largely dictated by the WiFi Alliance.
If your world is one where you are tasked to manage an on-premises network, wired or wireless, Cisco or otherwise, I have one other lesson I’d like to pass on. Network management is an important part of what you do and there was never the perfect product, not even from Cisco, that worked really well. Until now. That’s where I get to introduce Cisco Prime Infrastructure.: a product that is much less expensive than the competitors and quite powerful. You may not know this, but Cisco’s not had the best of luck in the network management department. (Maybe you do.) Cisco Prime Infrastructure, and the advancements made with our latest offering v2.2, changes that. It’s an on-premises solution that runs as a virtual appliance. It monitors all network elements not flagging up/down, rather health. It automatically manages configuration archives and it has a very robust set of troubleshooting and reporting tools. The topology map is surprisingly easy to use and understand. If you want to have it push out scheduled configurations, you can designate? that, too!
If you’re managing a network of more than 15 or 20 elements, consider Prime Infrastructure. If you want to kick the tires, you can download now and use it for 60 days and with 100 devices. Like it? Just add a permanent license.