This past week was spring break so I took advantage of the fact that there were no users around, and fiddled with Trapeze/Juniper’s RingMaster management software and the RRM configuration. This post will be an introduction to Juniper’s RRM, it’s defaults, and some options, and will lead to more posts on testing it. My network is a mixture of OG Trapeze MP models and new Juniper WLA models, mostly using auto-channel and default static power settings. RingMaster and most of the documentation I found is branded as Juniper.
Quick history lesson on Trapeze Networks:
- founded in 2002
- bought by Belden in July 2008 for $133 million
- acquired RTLS company Newbury Networks in December of that year
- was acquired by Juniper Networks in November of 2010 for $152 million
Juniper sort of splits its RRM into two areas: RF Auto-Tuning (power levels) and Adaptive Channel Planner (channel selection). I’m going to refer to these as RFAT and ACP. I say “sort of splits” because depending on whether you’re using RingMaster or the controller’s CLI for configuration, “auto-tune” can refer to power levels only, or power and channels. In general, I’ll only mention a CLI command if it performs a function not found in RingMaster. The docs that speak specifically about RFAT and ACP are from Trapeze and don’t seem to have been updated by Juniper. It looks like Ryan Adzima (@radzima) wasn’t far off with this tweet:
Trapeze says the power tuning calculation is done by the AP with no involvement from the controller. Access points listen for other APs operating on the same channel, and adjust their power to provide good coverage while avoiding CCI.
RFAT can be enabled in RingMaster in the Radio Profile configuration. In my current network configuration, each school has it’s own radio profile. Other settings that can be managed in RingMaster are:
- Power Tune Interval – time in seconds between power adjustment decisions
- Power Ramp Interval – time in seconds between 1 dBm adjustments when a decision to adjust the power has been made
- Minimum Power – the lowest allowed power in dBm that RFAT can select
- Maximum Power – the highest allowed power in dBm that RFAT can select
You can see the Radio Profiles Auto Tune tab here:
With respect to the Power Tune Interval, access points are constantly gathering data about the RF environment, and use that data to decide whether a change should be made when the interval ends. One thing I find strange is that, while you can configure an individual AP’s static transmit power level down to 0 dBm, the lowest allowable minimum for RFAT is 8 dBm. The Maximum Power option is set on a per-AP basis, as shown in the WLAN Access Point (WAP anyone?!) Properties 802.11na (or ng) Radio tab. The maximum allowed power level is 20 dBm.
There is no way to configure individual AP’s to use RFAT in RingMaster, but the “set ap power-mode“ command can do so at the controller’s CLI.
There is also a “set ap radio tx-power“ command. You can do the same thing in RingMaster, as you can see in the above image of the WLAN Access Point Properties. I am including it here to show how the default power level for access points is selected, and what the “high” option in the “set ap power-mode” command indicates.
The following table shows the highest statically configurable power level for each channel that I currently use.
I put this table together by selecting the channel in question in the WLAN Access Point Properties window and hovering over the Transmit Power adjustment arrows. The min and max values are then displayed in a pop-up as shown below.
Comparing my table with the description of the maximum levels in the “set ap radio tx-power” command raises some questions, as the levels don’t seem to match up. I would like to think that transmit power means transmit power and not EIRP.
Lastly, the command “set radio-profile auto-tune power-lockdown” can be used to set the automatically configured power levels to static.
Adaptive Channel Planner
ACP can be enabled for individual AP’s in RingMaster in the WLAN AP Properties 802.11ng(a) Radio tab. Just check that box! The first application of ACP, whether the AP is a new install or existing infrastructure, assigns a random starting channel. Trapeze says that quickly choosing a reasonable channel is better than waiting for the selection algorithm to perform at its next interval, and that reasonably good tuning usually results from the random channel selections. At CWAP class, the guys from other vendors said their APs will scan for a period of time before assigning channels. That sounds like a better implementation to me. Unlike RFAT, the controller is involved in ACP. At the end of every interval (one hour by default) an algorithm, distributed between the controller(s) and APs, evaluates the channel selection based on:
- amount of noise on the channel,
- packet retransmission count,
- utilization calculated on the number of multicast packets per second that a radio can send on a channel while continuously sending fixed-size frames over a period of time,
- PHY error count, and
- received CRC error count.
The channel will also change if severe interference is detected, regardless of the interval configuration. At a glance, ACP seems more complex than RFAT.
You can at least specify a list of channels for ACP to randomly choose from for both the 2.4 GHz and 5 GHz bands, as shown in the Radio Profile Properties window below.
That is pretty much all there is to see for ACP in RingMaster. There are a number of CLI commands that allow ACP configuration at the radio profile level rather than the AP level.
There is an “RF Auto Tune” window in RingMaster that allows you to set the day and time that the system actually evaluates the scanning data and makes tuning decisions. I can’t find any evidence that says this applies to power or channels or both. I would think that since you can set it to tune in the middle of the night, ignoring client connections could be the default.
This post will be a reference for me as I go through each school and try to clean up the hallway designs I’ve inherited. I’m thinking I’ll probably stay away from auto power tuning, but that of course could change. Most schools already use auto channel selection, so I’ll be having a good hard look at those configurations and see what happens. There will be some 2.4 GHz radios to turn off, and it’ll take a bit for the channel selection to react to that. I’ll be following up with more posts containing test results and new network configurations.
Thanks for reading and please comment! I’d love to hear of anyone else using Juniper and their experiences with it.
All CLI command images were snipped from this command reference. Note that link will take you directly to the 672-page pdf.