Coming off the Cisco Champion Radio podcast, François and Rowell share their RRM experience.
Our Cisco RRM Experience
We use Cisco RRM on a daily basis and in most of the environments we configure and manage. It’s come a long way since it’s introduction but we’re both still learning so much about Cisco RRM.
For a large environment, RRM is a useful tool. But when changes are done on the WLAN, RRM needs to be reset and settled.
In this episode, we talk about our experience using RRM. What does it mean to design with RRM in mind? How does one use RRM with high density, capacity, coverage, etc while keeping ensuring it meets requirements.
We touch upon using design tools such as Ekahau Site Survey and how you use those design tools with RRM. Can it be done?
And what do we think about those who don’t think RRM works at all? We touch on that just a little.
Take a listen to the episode and share your experience in the comments below.
François and I were fortunate to be able to host Cisco Champion Radio this week. We tackled the topic of Cisco Radio Resource Management (RRM) with subject matter expert, Jim Florwick.
Jim is a Technical Marketing Engineer with deep knowledge of how Cisco RRM works.
I started off the conversation by asking Jim what the most common misconceptions are of RRM. He provides a great response because many people seem to leave RRM at its defaults or believe RRM would fix all their RF problems.
Jim provides insight into how the algorithms work and how you should consider them into your design or existing WLAN infrastructure. The important tip he provided was to initiate a DCA restart when major changes have been applied to the WLAN. These changes can be new APs, channel changes, a new controller, you name it.
François asked an important question, how do you perform your validation surveys with RRM in mind? Jim responded with freezing RRM to prevent any changes from occurring while you’re conducting your validation survey. You don’t want to see the same AP shown with two different channels or have transmit power changing while you’re doing your survey.
We then go into the addition of FRA and the complexity it adds to RRM and other topics as well. Give the episode a listen on SoundCloud!
DCA, or Dynamic Channel Assignment, is a core component of Cisco Radio Resource Management (RRM). It runs a critical algorithm which dynamically changes an access points (AP) client serving channel based on multiple parameters, collectively called a Cost Metric. To really get an understanding of Cisco RRM, start with our previous episode on Cisco NDP.
DCA is configured under 802.11a/n/ac (5 GHz) and 802.11b/g/n (2.4 GHz)
Channel Assignment Method
The first thing you will notice is the Channel Assignment Method of Automatic, Freeze, or Off.
Automatic will run the DCA algorithm at the define Interval which is every 10 minutes by default. The anchor point is used to set the time of day the DCA algorithm would start.
For less frequent channel changes, increase the interval.
Keep in mind that clients will be disconnected briefly for a channel change.
Freeze does not run the DCA algorithm unless triggered.
Off disables the use of RRM and its auto-rf capabilities.
Avoid Foreign AP interference
Enabling this option allows RRM to take surrounding neighboring APs that are not part of your network into consideration.
RRM will create a channel plan to help mitigate co-channel interference.
Avoid Cisco AP Load
This option allows RRM to consider the traffic load on an access point to help build a better channel plan.
I don’t believe in enabling this feature because it can be difficult to make a client device associate with a specific AP.
Avoiding non-802.11a|b noise
This RRM option takes interference on the channel into consideration for the channel plan.
The category of interference is anything coming from a non-AP.
Persistent Device Avoidance
Sometimes there is non-Wi-Fi device interference affecting your wireless network. DCA can help mitigate the issue if Avoid Persistent Non-WiFi Interference option is enabled.
RRM will receive information from CleanAir and if the interference is creating a negative impact, RRM can issue a channel change to avoid it.
Viewing persistent devices can be done by viewing the details of an AP from the web interface. At the bottom of the page are all the persistent devices being tracked including the class type, on what channel the interference was heard, what the duty cycle is, RSSI, and when it was last seen.
DCA Channel Sensitivity
DCA has three different modes it could be in:
Scheduled – DCA algorithm runs at a selected time. Usually to minimize changes to off peak hours.
Steady State – DCA algorithm running at a specified interval.
Startup Mode – Used when making changes to the wireless network architecture. It restarts the DCA algorithm and runs for 100 minutes at high sensitivity.
By default, DCA has a medium channel sensitivity at 15 dB for 5 GHz and 10 dB for 2.4 GHz. This is the DCA Sensitivity Threshold.
By changing the channel sensitivity to high, then a cost metric of 5 dB better will recommend a channel change.
DCA Channel List
To have DCA dynamically select a channel, it must be in the Channel List. You can select specific channels for DCA to select from or select them all, including the UNII-2 channels.
Dynamic Bandwidth Selection
RRM has a flexible way to assign bandwidth to APs by analyzing the RF and selecting the best channel width.
By default, DCA uses 20 MHz channels. You can configure Cisco RRM DCA all the way up to 160 MHz channels or select Best to have RRM pick for you.
If an AP doesn’t support 40/80/160 MHz channels it will not be configured for it.
To see DCA dynamically select channels, head over to the CLI on the controller and enter the debug command:
debug airewave-director channel enable
*RRM-CLNT-5_0: Aug 27 12:04:48.270: 58:bc:27:33:33:33 Slot 1: Channel set request on 58:bc:27:33:33:33(1) to (100, 104, 0, 0) 40 MHz Before -62, After -72 Second best: ch 157 wd 40 mt -86 RRF:enabled(0) metric(199)
*RRM-CLNT-5_0: Aug 27 12:04:48.270: [spamGetMeshBhRRMStatus] 0
*RRM-CLNT-5_0: Aug 27 12:04:48.270: 58:bc:27:33:33:33 new channel width label set from 3 to 4
*RRM-CLNT-5_0: Aug 27 12:04:48.270: 58:bc:27:33:33:33 slot 1 Set new channel 100 width label 4
*RRM-CLNT-5_0: Aug 27 12:04:48.270: 58:bc:27:33:33:33 sent channel to spam and call trap(0)
*RRM-CLNT-5_0: Aug 27 12:04:48.270: 58:bc:27:33:33:33 Airewave Director: Final Channel Assignment (100,104,0,0) configured on AP 58:bc:27:33:33:33(1)
*RRM-CLNT-5_0: Aug 27 12:04:48.270: 58:bc:27:33:33:33 Airewave Director: set channel on cell 58:bc:27:33:33:33(1) to 100 from 100 to reduce co-channel interference from (-128,-128,-62) to (-128,-128,-72) reason 1
In the output above, Cisco RRM DCA made a request to change an AP to a 40 MHz channel width using primary channel 100 and secondary 104.
At the very end, RRM sets the channel to 100 with the reason – to reduce co-channel interference.
From command line, it is possible to view when the last channel assignment was set using the following show command:
show ap auto-rf 802.11a <ap-name>
Channel Assignment Information
Current Channel Average Energy............... -71 dBm
Previous Channel Average Energy.............. -71 dBm
Channel Change Count......................... 541
Last Channel Change Time..................... Sat Aug 27 12:14:59 2016
Recommended Best Channel..................... 116
802.11a – shows the 5 GHz radio 802.11b – shows the 2.4 GHz radio <ap-name> – replace with the name of your AP
The output was truncated just to display the channel assignment information. You can see the current and previous channel average energy, how many times the channel has changed, when it was last changed, and which is the best recommended channel.
Event Driven RRM
If you need the wireless network to react quickly and automatically to bad interference, ED-RRM is an option to enable.
An example could be a device using 100% duty cycle causing all nearby APs and clients to hold off communicating on the wireless medium. ED-RRM would allow an AP to make the channel change sooner than when the DCA interval kicks in.
To configure ED-RRM:
1. Click on Wireless
2. Click on 802.11a/n/ac or 802.11b/g/n
3. Click on DCA
4. At the bottom of the page, enable EDRRM
5. Set the sensitivity threshold to Low (my recommendation is not go above Medium)
The sensitivity threshold defines the air quality (AQ) at the following:
Low sensitivity – AQ at 35%
Medium sensitivity – AQ at 50%
High sensitivity – AQ at 60%
An air quality of 100% is good, very bad is 0%
The Rogue Duty-Cycle option will allow you to set the percentage of duty cycle and rogue duty cycle occurs before triggering EDRRM.
To learn more details about Cisco’s RRM – DCA be sure to check out the white paper from Cisco.
In this episode, we go over the Cisco Neighbor Discovery Protocol. The basis of Cisco RRM.
This episode is sponsored by Metageek.
Cisco WLC Neighbor Discovery Protocol (NDP)
Cisco NDP, short for Neighbor Discovery Protocol and Neighbor Discovery Packets, is a critical component of Cisco’s auto RF feature, Radio Resource Management (RRM). The purpose of NDP is to provide over the air (OTA) messages between access points (AP). It monitors and manages what each AP sees in the radio frequency (RF). It’s essentially how every AP sees other APs in an RF Group or Neighborhood. The end result is actual RF path loss between APs.
I see NDP as a way for APs to build a map of their locations in relation to each other based on RF propagation and path losses. Every 180 seconds (3 minutes), an AP will send an over the air (OTA) message to a multicast address, 01:0B:85:00:00:00, from each channel.
NDP messages are sent at the highest transmit power and at the lowest data rate supported for the channel being transmitted on. The transmit power and data rate selection is not configurable by the end user and is hard coded.
Cisco Neighbor Discovery Protocol forms the basis of many algorithms within Cisco RRM. Because of that, it goes without saying, if NDP doesn’t work neither does RRM.
Transmit Power Control (TPC – basis calculation for TPCv2)
Flexible Radio Architecture (FRA – basis for coverage overlap factor)
Rogue detection (If AP isn’t sending NDPs or unintelligible NDP then it is a rogue)
CleanAir (Used for interference reports)
CMX (For AP RF distance and path loss measurements)
As you can see, NDP is very important for RRM.
What’s inside a Cisco NDP Packet?
Slot ID for the sending radio
IP Address and Priority code of sends WLC
RF Group name converted to a hash for authentication
IP of sending AP’s RRM Group Leader
Are we using Encrypted NDP?
Version of NDP
Operating channel of the sending radio
Encryption Key Length
Encryption Key Name
Channel the NDP was sent on
The power (in dBm) the message was sent at
Antenna pattern of the sending radio
When an AP hears and receives an NDP message, it will validate the message from the transmitting AP to determine if it is a member of the same RF group. If it’s a valid NDP, the receiving AP will forward the message to the controller along with information such as the received channel and RSSI.
If the message is not in the same RF group it will be invalid and the packet will be dropped.
The message is added to a neighbor database. Each radio on each AP can store up to 34 neighbors ordered by RSSI, high to low.
There are two measurements taken:
RX neighbors – “How I hear other APs”
TX Neighbors – “How other APs hear me”
Configure Cisco NDP frequency
The frequency in which Neighbor Discovery Packets are sent out can be configured from the WLAN controller. By default, it is 180 seconds (3 minutes). It is recommended to keep this at the default.
The Channel Scan Interval is 180 seconds by default. Each channel dwell has to be completed within 180 seconds.
The Neighbor Timeout Factor is by default set to 5. This multiplier is multiplied by the Neighbor Packet Frequency value to come up with the timeout value. So with a default of 180 seconds for the Neighbor Packet Frequency x 5 = 900 seconds. This is the value used to determine when to prune access points from the neighbor list that have timed out.
If an AP were to disappear from the network, it would remain on other AP’s neighbor list until the pruning begins which is every 15 minutes.
To see these NDP packets over-the-air, I had two access points joined to my Cisco 2504 lab controller. Using Omnipeek, I set my adapter to to capture on Channel 64. I created a filter on the MAC address using addr(ethernet:’01:0B:85:00:00:00’).
In the screenshot below, notice the destination multicast address of 01:0B:85:00:00:00. Under the Relative Time column the NDP is sent out every 3 minutes, which is the default.
To find neighbor information using the Web GUI, click on the Monitor heading and on the left side, expand Access Points > Radios > Select radio.
Then move your cursor all the way to the right and hover over the blue dropdown box to select Detail.
Use Command Line
Using the CLI, we can view nearby APs from the controller by selecting which AP to get the view from. There are three options to select from. (Cisco Controller) >show ap auto-rf ?
802.11-abgn Display information for DualBand 802.11a/b/g/n.
802.11a Display information for 802.11a.
802.11b Display information for 802.11b/g.
802.11-abgn is used for APs with FRA.
802.11a displays information from the 5 GHz radio.
802.11b displays information from the 2.4 GHz radio.
Here’s truncated output from AP1 showing three neighboring APs. (Cisco Controller)> show ap auto-rf 802.11a AP1
AP 00:3a:7d:44:44:44 slot 1.................. -23 dBm on 36 20MHz (192.168.1.5) AP4
AP 58:bc:27:33:33:33 slot 1.................. -18 dBm on 100 20MHz (192.168.1.5) AP3
AP 58:bc:27:22:22:22 slot 1.................. -40 dBm on 44 20MHz (192.168.1.5) AP2
View NDP via CLI on AP
Another great debugging command is to view the RM measurements occurring from the access point. NDP packets will be sent out on each channel as you can see in the output below. I’ve truncated the rest of the messages. AP1#debug capwap rm measurements
CAPWAP RM Measurements display debugging is on
*Aug 23 18:17:46.016: CAPWAP_RM: Timer expiry
*Aug 23 18:17:46.016: CAPWAP_RM: Neighbor interval timer expired, slot 1, band 0
*Aug 23 18:17:46.016: CAPWAP_RM: Triggering neighbor request on ch index: 2
*Aug 23 18:17:46.016: CAPWAP_RM: Sending neighbor packet #2 on channel 44 with power 1 slot 1
*Aug 23 18:17:46.016: CAPWAP_RM: Scheduling next neighbor request on ch index: 3
*Aug 23 18:17:46.230: CAPWAP_RM: Notification for Request id: 4044, slot: 1, status 1
*Aug 23 18:17:46.230: CAPWAP_RM: Neighbor packet sent successfully on 44
*Aug 23 18:17:46.233: CAPWAP_RM: Notification for Request id: 4044, slot: 1, status 1
*Aug 23 18:17:46.233: CAPWAP_RM: Neighbor packet sent successfully on 44
*Aug 23 18:17:49.017: CAPWAP_RM: Timer expiry
*Aug 23 18:17:49.017: CAPWAP_RM: Neighbor interval timer expired, slot 1, band 0
*Aug 23 18:17:49.017: CAPWAP_RM: Triggering neighbor request on ch index: 3
*Aug 23 18:17:49.017: CAPWAP_RM: Sending neighbor packet #3 on channel 48 with power 1 slot 1
*Aug 23 18:17:49.017: CAPWAP_RM: Scheduling next neighbor request on ch index: 4
*Aug 23 18:17:49.159: CAPWAP_RM: Notification for Request id: 4048, slot: 1, status 1
*Aug 23 18:17:49.159: CAPWAP_RM: Neighbor packet sent successfully on 48
*Aug 23 18:17:49.162: CAPWAP_RM: Notification for Request id: 4048, slot: 1, status 1
*Aug 23 18:17:49.162: CAPWAP_RM: Neighbor packet sent successfully on 48
*Aug 23 18:17:52.018: CAPWAP_RM: Timer expiry
*Aug 23 18:17:52.018: CAPWAP_RM: Neighbor interval timer expired, slot 1, band 0
*Aug 23 18:17:52.018: CAPWAP_RM: Skipping neighor request chan 52; DFS channel
*Aug 23 18:17:52.018: CAPWAP_RM: Scheduling next neighbor request on ch index: 5
*Aug 23 18:17:53.327: CAPWAP_RM: Timer expiry
RM Neighbor debugging
Another debugging command at the AP level will display NDP packets being received from other APs, including those not in the same RF group.
AP1#debug capwap rm neighbor
CAPWAP RM Neighbor display debugging is on *Aug 23 18:31:33.529: LWAPP NEIGHBOR: Pak size 104 from 58bc.27xx.xxxx, interface - 1
*Aug 23 18:31:33.529: LWAPP NEIGHBOR: Updating existing neighbor 58bc.27xx.xxxx(1), rssi -35 on channel: 161 with encryption: 0
*Aug 23 18:31:33.532: LWAPP NEIGHBOR: NDP Rx: From 58bc.27xx.xxxx RSSI [raw:norm:avg]=[-36:-36:-36] [Neigh Srv Chan: Neigh Off Chan : NDP Pwr]=[161:157:20 dB] Rcv Ch Max Pwr [20 dB]
We briefly went over the purpose of Cisco NDP, neighbor discovery protocol, and how crucial it is to the RRM algorithms. I showed you where in the WLC GUI you can make changes to the NDP intervals but it is not recommended to change them other than the Neighbor Timeout Factor interval. In addition to capturing the NDP messages over-the-air, we saw how to capture NDP and some other RRM functions using debug commands from the controller and the AP.
The content of this post was first published on packet6.com
Robert Boardman is my special co-host for Episode 38. Robert is the creator of Hub Holster which is a great alternative to using velcro on your laptop to hold up a USB hub during a wireless site survey. I assisted in providing measurements for my own USB hub and after a few back and forth in communication I have a fully functional USB hub holder. It was printed using a 3D printer.
We discuss how the idea of the hub holster came to mind, the process that goes into building a hub holster, and other features we may want to see.
Both Robert Boardman and I will be heading to Cisco Live 2016 in Las Vegas. This will be my first time attending. I’m excited to meet others in networking and hope to get a podcast episode recorded with other attendees. If you are attending Cisco Live 2016, please let me know!
Our last topic of discussion is about RRM. Robert and I have been thinking about really learning the ins and outs of RRM. Our goal is to deploy RRM and collect the necessary metrics to determine whether or not RRM helped in our scenario.