Auditorium Density/Capacity Planning for Wi-Fi

I was recently tasked to do a design for a small 450-seat auditorium and provide capacity and throughput numbers. Those who have known me for a while probably know that this type of auditorium is kind of a sweet spot for me, having done designs for a number of church sanctuaries of various sizes. In this post, I’m going to get into the nitty gritty details of making sure that not only does an auditorium have sufficient wireless capacity to meet the connectivity needs of the space, but also to have realistic expectations of what the performance will look like in order to build sufficient backend networking infrastructure without needlessly overbuilding it.

Auditorium design should be simple, right? Here’s how I have seen it done, way too many times to count:

  • Count up how many seats there are, divide by some number of seats per AP (usually based on the AP data sheet), and then figure out how many APs that gets you.
  • Figure out your capacity by taking the AP throughput (again from the data sheet) and multiplying that by the number of APs. Then divide that capacity so you know how much bandwidth you get per person.
  • Try to do a predictive model using Ekahau, to place the APs in exactly the right spot, and without ever surveying the space.

So let’s say you have a 1000-seat sanctuary where you want to use a Ubiquiti Unifi HD access point because that’s what your colleagues on social media recommended. The vendor data sheet says that you can do 500 concurrent clients per AP, so that means two APs (let’s say three just for redundancy), and each AP can do 2533 Mbps . So you should be able to get 7.6 Gbps, divided by a thousand seats, which gives you 7.6 Mbps per client, and you’ll need a 10 Gbps switch. Easy job, under a thousand bucks for the gear. And then when you fill the room up, the whole thing collapses, everyone is complaining about how it doesn’t work, and you’re left wondering why.

Because that’s not how any of this works.

For starters, never believe the data sheet. That’s marketing, not engineering. There is no amount of marketing copy that can ever overcome the fundamental laws of physics. So let’s pick this design apart, piece by piece… (yes, I’m gonna pick on Ubiquiti for a bit here, because their UniFi brand is often thrown about as a solution to all your wireless problems by people who don’t actually understand how wifi works – but these principles apply to any vendor – no vendor has a magic bullet, you still have to do the engineering)

Caution: Math (or at least arithmetic) ahead. Don’t say I didn’t warn you. Hope you paid attention in school.

The Engineering

doin it rong:

Error #1: AP Throughput

This is probably one of the most egregious attempts by the marketing department to ignore reality. This number published on the data sheet (and also frequently wielded by consumer AP marketing) is completely bogus, but marketing loves to show off big numbers. It is typically created by taking the maximum possibly PHY rate (more on that in a second) on each radio, and adding them together. (why? you can’t aggregate client radios like that!). The number “2533 Mbps” comes from adding the max PHY on 5GHz (1733 Mbps) with the max PHY on 2.4 GHz (800 Mbps)

What is the PHY rate?

It is the speed at which an individual wireless frame is transmitted over the air. It can vary from one frame to the next, one client to the next, and is highly dependent on RF conditions. What goes into the PHY:

  • Channel Width
  • Number of MIMO Spatial Streams
  • Guard Interval
  • Modulation and Coding Scheme (MCS)
  • Resource Unit Size (in 802.11ax)

A table of all possible PHY rates (and the math behind them) can be found at the ever-handy mcsindex.com.

And here’s where this speed number comes flying apart. In order to achieve this maximum PHY, you need to use an 80 MHz channel (40MHz on 2.4 GHz, which is a monumentally bad idea), a short guard interval, 256QAM with 5/6 coding (which typically requires signal:noise ratio of over 40dB to achieve), and FOUR spatial streams. Given that the vast majority of devices in the wild only support two spatial streams (and the only 4SS client device is a desktop card), it’s safe to say that you’re never going to even come close to that maximum PHY rate. And even then, wireless is a half-duplex shared medium where only one device can talk on a channel at a given time. So even if you were to somehow get that max PHY, your throughput for a single device might be about half that at best. And as you add more clients, it gets even lower. Remember: Every TCP segment results in FOUR transmissions on the wireless: The segment itself, the layer 2 acknowledgement of that frame, then the TCP acknowledgement, and then the layer 2 ACK of the Layer 3 ACK.

Error #2: Constrained Resources

The most important thing to remember when doing dense Wi-Fi deployments is that your most constrained resource is not bandwidth, it’s airtime (the amount of time a given device gets to send data). In order to maximize airtime sharing, you want devices to get on, say their piece as fast as possible, and get off. This also means you want them to use as little spectrum as possible to do so. The key to supporting more client devices is to minimize their use of spectrum and maximize spectrum reuse (where multiple access points use the same frequency in a way that they don’t interfere with each other, which is a lot harder than it sounds)

Ultimately, the only way you can add capacity to a space is to add spectrum. I’ll demonstrate in a minute how channel width matters a lot less than one might expect.

And let’s not forget that while this AP advertises throughput of 2533Mbps, it only has a 1Gbps port to connect to the switch…

Error #3: Assumptions

We’ve probably all heard the old saw about what happens when you assume something. It still holds true in wireless engineering. An auditorium may have a thousand seats, but it’s also vitally important to understand how that space is used, what kinds of devices there are, how many people, etc. Broadly speaking, an auditorium will “feel” packed and completely full when there are about two thirds of the seats occupied. But if you’re selling reserved tickets, it’s entirely possible to fill every one of those seats. And what devices are those people bringing? There’s a big difference between a 1000-seat auditorium that has 700 people in it for weekly worship and when that same space has 500 people in it attending a conference, or when 1000 people are there watching a film or a performance. Ultimately you want to plan around the most likely intensive usage scenario, which is going to typically be a conference (although I’ve done plans that assume the most intensive scenario is something completely insane like an Apple product launch).

Planning (Doing it right)

So let’s run the numbers for this fictitious auditorium that seats a thousand people. broadly speaking, this room is going to be of such a size that no matter where you place the AP, it’s going to light up the whole room. At this size, you’re not going to get any frequency reuse, even with directional antennas. If you were hoping to use the crowd to attenuate the signals and get reuse that way while putting your access points under the seats, stop now – Aruba (who have tested and deployed a whole lot of venues of all sizes) do not recommend going under the seat in any venues under about 10,000 seats unless you simply don’t have a means to go overhead.

Since we’re not getting any channel reuse, this gives us a grand total of 500 MHz of spectrum to work with, plus another 60 MHz in the 2.4GHz band – but it’s probably best to simply forget about 2.4 GHz in an auditorium because a bunch of A/V stuff is using it (and likely ill behaved stuff at that), not to mention the hundreds of wearables the people in the seats have, which will light up the entire Bluetooth channel space. So let’s go with 5 GHz for now. I’ll talk about 6GHz later.

In the 5 GHz band, we have:

  • 25 channels at 20 MHz (500 MHz)
  • 12 channels at 40 MHz (480 MHz)
  • 6 channels at 80 MHz (480 MHz)
  • 2 channels at 160 MHz (320 MHz)
5 GHz Channel Allocation (Credit: Jennifer Minella, SecurityUncorked.com)

I’m gonna go ahead and say it: Don’t waste your time with 160MHz. Sure, you get some sick PHY rates with it, but device support is limited. And don’t forget that weather radar can remove 3 channels at 20 MHz, 2 channels at 40 MHz, and 1 channel at both 80 and 160 MHz – but unless you’re very near a radar site, and the radar is penetrating from outside, you can use these channels without any issue. I’ve even seen these used inside airport terminals within view of the TDWR. Use these channels right up until you can’t.

So how do you choose what channel width to use? The only difference is whether you have more devices talking at once, at lower speeds, or fewer talking at once, but doing so at higher speeds. In the end, it doesn’t make that much of a difference to your throughput, and then it becomes a decision of how many APs you can physically put in the space (and their specific placement in a small auditorium is not too picky, since every AP lights up the entire space). 12 APs is a good flexible middle ground here, because you can do 12x40MHz channels. or 24x20MHz if the AP supports dual 5GHz radios (such as the Aruba AP-340 or AP-550 series access points), or 6x80MHz and leave the other 6 as spectrum monitors. Or adapt as needed.

Let’s now plan on a full conference load of 500 people, who each brought a laptop, a smartphone, and a tablet. and will be evenly distributed throughout the room (because elbow room and personal space). The tablet and the phone will be doing typical low-usage background stuff while the laptop will be doing much heavier usage, let’s say 1 gigabyte per hour (which is roughly equivalent to a 2Mbps video stream – I’m thinking this is something like the Church IT Network conference and they’re all geeks doing geek stuff), and that about 3/4 of them are active, the rest have shut their devices off to minimize distraction. I’m also going to plan on these being 2SS MIMO devices, since that’s the overwhelming majority of what’s out there.

So here’s the breakdown, assuming most clients link up at MCS7 with a standard Gaussian distribution on either side. We’re also assuming a 50% net ratio of usable throughput (goodput) to PHY speed. Duty cycle is how much of the available airtime is used for this load – you want to try and stay under about 60% to accommodate for neighbor interference, etc. Much above that and performance really starts to suffer. These calculations are based on an excel sheet that I have, but it’s a little rough around the edges, so I haven’t shared it here.

24x20MHz12x40MHz6x80MHz
Devices113011301130
GB/Hour400400400
Available Throughput156016201755
Duty Cycle57%55%50%
Average Throughput per client1.38 Mbps1.43 Mbps1.55 Mbps

And this is where things get a bit counterintuitive (as they often do with Wi-Fi): You’re slightly better off here going with fewer APs at 80 MHz than you are with more APs at 20 MHz – but if you lose an AP or a channel due to failure or radar hit, you lose a lot more capacity when using the wider channels. In any case, you can see that all you actually need for this room is a gigabit switch with a 10G uplink, and a decently fat pipe to the internet. You also need at least a /21 IP address space (but probably a good idea to go to /20 or even /19 to accommodate for MAC randomization). You also want to plan on sufficient AP capacity outside the space for devices to transition to during breaks and whatnot, but they won’t need nearly as much airtime capacity as those devices are not going to be using it as heavily as the laptops.

The Math

Input data:
  • Infrastructure:
    • Area Population (Head Count) – the number of people in the room. Distribution Curve: Normal/Gaussian
    • Number of access points (self-explanatory)
    • Channel Width (2.4GHz, 5 GHz) (Not directly used in calculations, only in determining link speed input)
  • Client Devices:
    • Wifi Devices per person (Distribution: triangular)
    • Gross Take Rate (how many people using wifi (Gaussian)
    • % Devices on 5GHz (if using both bands)
  • Client Activity Modes: (activity per hour, in MB)
    • High/Medium/Low (Gaussian)
  • Activity Distribution (percentage of traffic in each mode, Gaussian)
  • Link Parameters (I shoot for the MCS7 values on 2SS – but what you can realistically expect will also be a function of how far the AP is from the seats, which is a factor in tall rooms):
    • 2.4 GHz Link Speed (Mbps, median speed, triangular)
    • 5 GHz Link Speed (Mbps, median speed, triangular)
    • TCP Net ratio (Goodput/Link speed, triangular)
Distribution Curves: a) Normal/Gaussian, b)Rectangular/Uniform, c) Triangular/Continuous, d) U-shaped/quadratic
Output Data:
  • Connected Devices: Headcount * Devices per person * Take Rate
  • Client Demand (MB/hr): (Sum of: (activity mode * activity percentage)) * headcount
  • Available Throughput (Mb/sec): AP count * Link Speed * Goodput Ratio
  • Duty Cycle: ((Client Demand * 8)/3600) / Available Throughput

You’ll also want to apply the distribution curves to all those values to establish your 95% confidence ranges. Hit me up if you want details..

You can also improve your airtime efficiency by narrowing the range of PHY speeds so as to keep extra slow clients from connecting and chewing up your airtime – This is accomplished by setting your basic and available data rates to a higher value such as 12 Mbps or 24 Mbps. Also, don’t forget that because any slice of airtime is at a premium, don’t go crazy with your SSIDs, to keep your beacon overhead under control even at the higher basic rates. You also don’t want to “hide” any SSIDs in order to keep your unassociated clients from chewing up airtime with probe requests that are trying to figure out if the hidden SSID is one they know about. You want as many devices in the room as you can get to associate to something, anything and shut up with the probes already. Even if it’s an open SSID that goes nowhere.

Caveats

It is worth noting here that artificially throttling client speeds will do more harm than good – the additional traffic overhead that comes with that eats up airtime like crazy. So don’t see this and think you should limit your client devices to 2Mbps in order to make sure the system doesn’t get overwhelmed – see Jim Palmer’s presentation “The Netflix Effect on Guest Wi-Fi” for why throttling client speeds doesn’t work the way you think it does.

These calculations also doesn’t factor in any airtime overhead from adjacent APs outside the space, which is one reason why you want to keep your airtime duty cycle under 60% and your goodput ratio to 50%. Once the system is deployed, you’ll want to validate in the field what they actually look like, which will give you a good idea of actual usage and how well the model predicted your capacity.

What about placement and directional antennas?

In an auditorium this size, it really doesn’t matter. Because no matter where the AP is or what antenna it has on it, it will light up the entire room, even at a low power setting like 10dBm. Don’t get me wrong, I’m a huge fan of using directional antennas to sculpt the RF footprint. But unless you’re dealing with a small stadium, you’re not going to get frequency reuse out of directional antennas anyway (and a directional antenna can actually cause you more trouble – if the hot spot of the signal is too narrow, even way off-axis you’ll still be above the -82dBm contention backoff threshold in most of the room due to reflections and how focused your antenna is). If you want a good visual of this, go find one of the lighting people and ask them to aim a lighting fixture with a narrow beam at a seating area, turn on only that light, and vary the brightness… You’ll get enough scattered light in most of the room to see where you’re going. Light is, after all, still electromagnetic energy, so your RF is going to behave in similar ways.

Because the APs light up the whole room, you can literally put them anywhere that’s convenient for installation or maintenance access (just don’t put them too close to each other). There are however some cases where you can (and probably should) use a directional antenna in an auditorium space:

Tall ceilings – if you’re stuck with mounting the APs on a ceiling that’s much more than about 10m from the seating area, use a directional – at that height, 90° is still going to cover the entire floor, and 60° likely will too (remember that antenna beam width is considered to be between the -3dB points on the antenna plot, and in a space like an auditorium, your functional beam width is going to be closer to between the -10dB points, and you’re going to get a lot of scatter from the back lobes of the antenna as well, something that Ekahau doesn’t model – but this multipath environment can ultimately help with MIMO.

Keeping the signal inside and the noise outside – this is another place where you might consider directional antennas – if your APs are near the perimeter of the space and there’s space outside that also has Wi-Fi, a directional antenna can keep the outside signals from causing contention, as well as keep the signal from spilling into the area outside and causing contention with the APs external to the room. It’s also probably a good idea when you’re building a new auditorium to build the shell of the room such that it has high attenuation between the outside and inside (tilt-up precast concrete panels are great for this, but there’s a case to be made for intentionally designing RF shielding into the walls. It probably doesn’t hurt to set the room to a different BSS color if you’re using 802.11ax – but I haven’t yet encountered this in the wild. Last year, I was working from someone else’s design in a cruise ship where there were no fewer than 40 APs in the ship’s theatre, which seated 750. These APs were not only using a 60° directional antenna, it was placed immediately behind an expanded metal mesh used to support acoustic treatment fabric. And yet even at the lowest power I thought I could get away with, that one AP was still lighting up the seats below (about 6m) at -60dBm… The back lobe of these antennas was bouncing off the steel structure of the ship, and the weakest spot in the room was directly on the center axis of the directional antenna. I ended up putting most of those APs in spectrum monitoring mode, and making notes for the next ship auditorium. Upside is that a steel ship gets GREAT frequency reuse elsewhere.

Aesthetics – Sometimes you just want to hide the APs – and in that situation, an external antenna can be easier to hide than a whole AP. But also bear in mind that most APs now also have BLE functionality, and the BLE antennas are still inside the APs even if the Wi-Fi antenna is external. So if BLE is a design consideration, keep that in mind. You can also hide APs (or antennas) by skinning them (printable automotive vinyl wrap is great for this), painting them (if the manufacturer allows this, just make sure you use nonmetallic paint), a paintable cover (Aruba offers matching paintable covers for almost all of its indoor APs) – I haven’t tried it, but I wouldn’t be surprised if you could also hydro-dip the covers or the radomes. You can also hide APs in an enclosure such as the Oberon 1019-RM or otherwise camouflage them (See previous post: Hiding In Plain Sight). But one thing you don’t want to do is put them all being the acoustic panels where they all have line of sight to each other, as this will screw with 802.11k as well as automatic channel/power algorithms like AirMatch. This is the same as putting your APs above the ceiling tiles.

What about 802.11ax?

802.11ax (“WiFi 6”) brings a few airtime efficiencies to the table, but that will mostly manifest itself with the low traffic clients that don’t need to use the full data payload of a frame. High traffic clients will typically use all the RUs available in a single transmission, so our airtime usage calculations should not assume any OFDMA gains. BSS coloring (see above) may also be useful.

What about MU-MIMO?

Even if you have devices that support it (rare in 802.11ac, required for 802.11ax), MU-MIMO frames don’t really happen all that often in the real world, so planning your capacity around being able to use it is not a great idea. If you can somehow get MU-MIMO, then you’ll see some more efficient airtime usage. Again, we can’t count on this, so our capacity calculations should assume it isn’t happening.

What about 6 GHz?

6 GHz is pretty simple – you get to add more lots more spectrum, which directly translates to more capacity/throughput. It seems likely at this point that most vendors will release some kind of tri-radio/tri-band access point that will simply add the ability to run a 6 GHz channel, so you would simply calculate the additional capacity as additional APs and swap them out when the APs become available. But also consider that client support may not be fully available for a few years, so when you run your calculations, do them for 5GHz only and then treat 6GHz as a supplemental capability. If you’re running a dozen 5 GHz APs with 40 MHz channels, you can use those same 12 APs with 80 MHz channels on 6 GHz and the higher throughput alone should encourage any 6 GHz capable client device to choose the 6 GHz connection. Band steering without the band steering.

6GHz Wi-Fi Spectrum (Image Credit: Wireless LAN Professionals)

What about 2.4 GHz?

Leave it. Pretend it doesn’t exist. An auditorium full of people is going to be chock full of Bluetooth signals from wearables and wireless earphones (not to mention an increasing number of hearing aids). There’s also a lot of A/V stuff that lives in 2.4 that you just don’t want to worry about either. If you’re unable to convince the theatrical engineers to integrate with your existing infrastructure, you may also want to leave one 20MHz channel on 5GHz for them (165 is easy). And you only gain 60 MHz of spectrum, at the expense of a lot of headache.

tl;dr

Planning your auditorium capacity isn’t just a matter of taking the vendor specs and multiplying it by a certain number of APs per seat. There’s much more detailed engineering and calculation involved, and if it’s not something you’re comfortable doing or you don’t understand the numbers, hire a pro who can do the engineering for you – it’s going to be a lot cheaper than buying the wrong thing several times over…

Additional Resources

Props and Shout-Outs

Thanks to the following people who contributed their expertise and knowledge to this post:

Aruba AP Provisioning

As part of trying to wrap my own head around the various profile dependencies in actually provisioning an Aruba AP , I’ve mapped it out. This is the <stuff> that goes into this process:

provision-ap
read-bootinfo {wired-mac|ip-addr|ap-name} <data>
<stuff>
reprovision {serial|wired-mac|ip-addr|ap-name} <data>

As you go to provision an AP, start on the outside of this map and work your way in. This will make sure that all the various profiles you need are in place. The web UI hides some of this stuff from you and doesn’t organize it as logically as one might expect.

When doing this on the CLI in Mobility Master Conductor, make sure you’re in the right corner of your hierarchy (namely, /md or /md/GROUP). And remember that on MMMCR, show run is not nearly as useful as show config effective… And config purge-pending sure comes in handy when you goof something up.

You can also do show profile-hierarchy but that only shows the profile entries… And it doesn’t fit neatly in a terminal window…

Caveat: This is not comprehensive by any stretch. There are dozens more options, these are just the more common ones. If I goofed, let me know. All the gory details can be found in the ArubaOS User Guide.

“It’s ALWAYS DNS (or DHCP)”

There’s a common saying among my network engineering peers: “It’s ALWAYS DNS!”. For those not familiar with the concept, this refers to the alarming regularity with which networking troubles end up being caused by something trivial, such as name resolution. And when it’s not DNS, it’s usually DHCP. Those two troublemakers alone are responsible for some ridiculously large percentage of network support issues. (At least until someone at a tier 1 provider inserts a typo into a route table advertised to half the internet via BGP, and takes everything down, but I digress.)

Last weekend, I rebuilt my home wireless network from an Aruba Instant cluster back to a controller based network, using ClearPass as an authentication and authorization backend for the home network. Gross overkill for a home network, but it gives me stick time on stuff that I need to know for work, at a much grander scale.

But first, a little background into the Aruba Way of doing things: In an Instant cluster, the wireless networks are bridged to a VLAN that is trunked to the access point. You can also do this with campus networking, but managing all those VLANs on every port that feeds an access point is usually a recipe for forgetting something vital. So the campus model lets you build a single access VLAN on your AP ports, and the AP establishes a GRE tunnel back to the controller cluster (which also allows for some great redundancy and high availability options), and the various VLANs terminate on the user anchor controllers (because each user has their own tunnel back to the controller, which allows you to segment their traffic out and handle it at layers 4-7 based on a variety of rules, and the only thing going over the wire is an encrypted tunnel, which is a significantly better security posture should someone unethically decide to monitor traffic on a switch port when they shouldn’t.

This is also where ClearPass comes into play – How user sessions and traffic are handled is defined in roles. Each role consists of various rules. How roles are applied are defined by policies. You can map roles to users and/or machines with the magic of ClearPass, and then when someone connects to the wireless network, ClearPass can return a role (and it can map a different role based on whether you authenticated with a username/password, a certificate, or any one of a number of other data inputs). Basically, when ClearPass returns the OK to the controller, it also includes a bunch of attributes for that user, including roles. It’s extremely powerful magic, and when wielded wrong, it can cause no end of heartache trying to figure out just what exactly went haywire. And I’m still very much a ClearPass n00b.

Which brings me back to my newly built and ClearPass-enabled network. And so like every good story…

No $#!+, there I was…

When I connected, it would take a good 10 minutes before I could access the internet. And so, I’m wondering what I screwed up in my ClearPass setup that would have done this… But the roles were being assigned correctly, and the rules associated with those roles were pretty straightforward: “allow all”. So why in the heck were devices on the home network taking forever and a week to get an address? This was not happening on my IoT and guest networks.

First, I realized that my devices were associating just fine, so ClearPass and the role derivation were working correctly, which immediately acquitted the Wi-Fi (but as far as the others in my house were concerned, the Wi-Fi was still screwed up). But that meant I had a good Layer 2 connection. I tried to make sure that the VLAN was properly connected from the pfSense router to the core switch, and the controllers (running in VMWare) were properly trunking to the distributed vSwitch and also out to the core switch. Everything on that front looked good. I tried manually assigning IPs to the wireless clients on the home LAN, and they worked great. So L3 worked, which implied L2 did as well. And when clients on the home network did eventually get an IP address, they worked fine as well. So nothing was being bottlenecked anywhere either (I should hope not, as the VMWare hosts and the router are all connected to the core switch with dual 10-gigabit fiber links!).

After a few days of racking my brain over this, and hearing the people who live in my house continue to complain about network weirdness (thankfully, my family is not doing virtual school/work… except for me), I finally resigned myself to doing what I should have done in the first place: Breaking out Wireshark and figure out just what was actually happening on the network. DHCP is pretty simple, so finding out what broke should be straightforward, right?

Quick refresher on DHCP: The process of obtaining a DHCP address goes like this:

Since I knew I had good L2 connectivity, I fired up Wireshark on my laptop, capturing what was going on at L2, and would move to other points in the network if I needed to. The first thing I saw is that a residential network, even with isolated guest and IoT traffic, while nobody else in the house is using it, is a fairly chatty place. I saw a bunch of multicast traffic (I have a lot of Apple devices), even IP broadcast traffic. And there, among all that, was the DHCP process. Discover. Discover. Discover. Offer. Request. Request. Request. Discover. Discover. Offer. Request. Request. Request. Discover. Discover. Discover. Offer. Request. Request. Request. The more astute among you may have noticed something missing from this sequence. Something rather… important.

Turns out, my DHCP server was making an offer, and then ghosting my devices as soon as they responded to that offer. And periodically, a DHCP ACK would sneak through. And by now, it had started happening on my IoT network as well, as half my Nest Protect alarms were now showing offline. But that told me one very important thing: that my DHCP server was in fact online, reachable, and responding. Up until that very last point.

So I then did what any sane engineer would do:

I had already restarted the dhcpd on my pfSense box, so I didn’t have much faith in the curative effects of a digital boot to the head, but what the heck, can’t hurt, right?

And that’s when I saw it. I went down to my lab, and there, on the front of the DL360 that is running my router, is an angry orange light which should normally be a happy little green. Uh-oh.

So, I pop out the handy little SID tray, to see what it’s angry about… And this is not something a server admin wants to see:

Yep, that’s flagging all three memory modules in Processor 1’s Bank A. This just became more than a simple reboot. Sure enough, when it went through POST, it flagged all three modules. Power off, slide out the server (rails FTW), and perform that tried and true troubleshooting method I learned and perfected in the Air Force a quarter century ago: Swaptronics. Move a suspected bad component and see if the problem follows. So, I switched all the DIMMs from bank A with those in Bank B. If the fault stayed with Bank A, then I had a bum system board. If the fault followed the DIMMs to Bank B, then the fault was in the DIMMs. I really wanted the fault to follow the DIMMs.

Plug it all back in, and fire it up, and the fault was…

NOW IN BANK B!!!! Hallelujah, I don’t have a bad server on my hands!

So now I shut it down, tossed the bad DIMMs in the recycling bin (yes, our recycling pickup actually takes e-waste, which is really nice when you’re a nerd with way too many electronic bits), and repopulated/balanced the banks (I also had to remove a fourth DIMM to keep things even, but it’s a known good part, so it did not go to recycling).

I fire the machine back up, and yay, it’s no longer grumpy about the bad memory, although it is briefly perplexed by the fact that it now only has 24GB instead of 32GB, and has somehow realized that it just had a partial lobotomy. After a few minutes of much more intensive self-testing than usual, it boots up pfSense, and gives me the happy beeps that pfSense does when it’s fully booted (for those of us who run our pfSense boxes headless!)

The moment of truth: I connect my laptop to the Wi-Fi (with the wireshark still circling)… and sure enough, the DHCP ACK comes through on the first try… So as near as I can tell, whatever part of the system RAM contained the bit of code required to send the DHCP ACK had suffered some kind of stroke, but not one severe enough to take the whole box or even the operating system down.

See? It’s always DHCP.

EDIT: Turns out there was also more to this – Wired clients (and access points) started getting DHCP right away after fixing this, but wireless was still giving me fits. As it turned out, There was something about the Aruba mobility controllers terminating user sessions that played havoc with the hashing algorithms that VMWare uses to handle NIC teaming on switch uplinks, and the ACKs were coming back through a different path and getting lost along the way.

For the moment, I disabled one of the 10G links to the switch until I can figure out what magic incantations I need to make on the vSwitch to get the hashing algorithms to properly use the multiple connections with the VMCs – or I may just use the second 10G interfaces for vMotion or something.

and that, kids, is how I used Wireshark to diagnose a system memory problem.

Hands On : Aruba Instant

After our quick little tour of Aruba InstantON, I’m going to move up to the next level of Aruba gear: Instant.

The naming can be a little confusing to the ArubaNoob, but Instant has been part of Aruba’s product offering for a very long time. While it appears controllerless, it still makes use of a virtual controller that lives inside the APs on the network (and in case the AP running the controller goes offline, the remaining APs on the network decide on a new leader by holding a rap battle or a dance-off. OK, just kidding. They actually do a sort of digital version of Rock, Paper, Scissors, Lizard, Spock.

This virtual controller concept has also been done by Ruckus with their Unleashed platform, which in terms of functionality is somewhere between Instant and InstantON, and Cisco’s Mobility Express. I’m not 100% sure, but I think Aruba had it first.

In previous generations of Aruba access points, you either purchased an Instant AP (IAP), a Campus AP (CAP) , or a Remote AP (RAP). The latter two required a Mobility Controller (MC). You definitely couldn’t RAP without an MC. Now, all APs ship as Universal APs and figure out which mode to be when they boot up, and can be easily converted from one to the other (in the dog park that is Ruckus Unleashed, you would have to reimage the AP with new firmware).

Who it’s meant for

Instant is designed for small and medium business environments, and home labs of geeks who subscribe to the idea of “if it’s worth doing, it’s worth overdoing” (My home wireless network right now consists of 7 APs in an Instant cluster). It also is very useful in large enterprises that consist of many small locations, especially once you start managing them all with Central. If you have a chain of coffee shops or boutiques that only require a few APs, then Instant+Central is definitely something you should look at. If you only have one, InstantON is more your speed.

Instant does not require any per-AP licensing, but it still includes a lot of the features you find on the campus systems. It even includes an internal RADIUS server and user database so you can do enterprise authentication (as of 8.7 which was just released in July 2020, you can even do up to 24 unique passphrases with MPSK before having to get ClearPass involved, which is real handy for IoT networks that use crappy chipsets that don’t support enterprise auth). It will also do an internal captive portal. It still has role-based access control, which provides layer 3 policy enforcement at the AP, including content filtering. And much like the InstantON APs can do, you can even use an Instant AP as your internet gateway (guess where InstantON learned it from?). You can even use it with ClearPass and all the goodies that come with that.

When a Universal AP powers up, it goes through the following process:

If setup mode is not accessed within a period of 15 minutes, the UAP reboots and goes through the process again. It can be a lonely existence. (this mode is not unusual to find in large campus networks where there exists a network disconnect at Layer 2 or Layer 3 between the AP and the controller. Chasing these down on a cruise ship is maddening… but it gets you a lot of steps.)

Setup Mode

Once the AP is in setup mode, it will broadcast an open SSID called SetMeUp-DD:BE:EF (where the last half is the last half of the wired MAC address of the AP). Connecting to this SSID will bring you to the configuration page (it will even conveniently pop it up in the captive portal window if your OS has such a thing). You can also access this by opening a browser to https://setmeup.arubanetworks.com, which it looks up via mDNS. (Caveat: This doesn’t work so great if the AP does not have an uplink and an IP address on the network, even if that IP is not routable… And accessing it via IP address only redirects to the hostname, and mDNS doesn’t really like not having a network to do its thing. So give it an uplink, even if it’s just a WLANpi.)

I once was traveling through a midwestern airport where I was scanning the wifi (it’s a wifi nerd thing) when I saw a lone AP broadcasting “Instant” (which is what Instant used to do before AOS 8.x). I eventually found the AP in a restaurant, where it was sitting all by itself on the ceiling, still in setup mode with the defaults… A quick peek into the setup page showed that this thing had never been configured… I found the manager to let them know that someone didn’t finish a job they were likely paid handsomely for, and she told me it had been there for almost 3 years and nobody had any idea what it was for or remembered who installed it or when. The airport’s installed public system was Meraki.

Once you’re in the setup interface, you can then configure it to your heart’s content. Then, when you bring up a second and subsequent access points on the network, they will find the first one, grab the configuration, and join the party. This scales surprisingly well – you can run several dozen access points on a network like this (There’s no actual hard limit, and it’s been officially tested up to 128 APs, but this is definitely not recommended – that’s well into Campus AP territory). It may not be truly instantaneous (we do love instant gratification), but it’s pretty darn close.

Limitations

There are a few limitations to this mode of operation, in addition to the aforementioned scaling issues (if you’re used to a SOHO/SMB system like Ubiquiti, 100 APs will sound like a lot to you. Once you get into controller based networks with Aruba, even a thousand APs is middle of the road – I routinely work with networks well in excess of this).

A few of the things you can’t do with Instant:

  • AP Groups
  • AirMatch (Instant uses the older ARM techniques for RF management)
  • Tunneling to controller (yet…)
  • I’m probably forgetting some things…

Perhaps the most useful aspect of Instant is that it can either be managed in the cloud with Aruba Central (if you’re used to Meraki, you’ll love Central), or if your network requirements grow to where you need to get a controller involved, switching the APs over to that mode is quick and easy, and you don’t have to buy new gear.

Labbing It Up

If you want to play around with Instant, it’s pretty easy: Buy an AP. Or more. If you have to fund your own lab gear, there’s a ton of used and refurbished Aruba gear on Amazon or eBay (If you go with HPE Renew, you still get HPE’s legendary lifetime warranty on network equipment). Recently, I saw a whole bunch of Renewed AP-345s on ebay for under $200. Just make sure you get the correct country code (US or RW) – the two can’t coexist on the same Instant cluster (in a controller environment, the controller country code takes over and ignores the AP setting).

If you’re new to the Aruba product line, here’s a quick cheat sheet to figure out what kind of AP you’re getting. It’s not 100% exact, but it should give you a general idea of what you should be getting.

The first digit of the 3-digit model number indicates product generation:

  • AP-0XX (or just AP-XX): 802.11g
  • AP-1XX: 802.11n
  • AP-2XX: 802.11ac Wave 1
  • AP-3XX: 802.11ac Wave 2 with integrated BLE
  • AP-5XX: 802.11ax with integrated BLE and ZigBee

The second digit indicates capabilities (1XX series and up)

  • AP-X0X: 2 spatial streams
  • AP-X1X: 3 spatial streams (although the 51X series is 2SS on 2.4GHz and 4SS on 5GHz)
  • AP-X2X: 3 spatial streams, second Ethernet port
  • AP-X3X: 4 spatial streams, SmartRate port, Gigabit Port
  • AP-X4X: 4 spatial streams, dual SmartRate ports, dual-5GHz,
  • AP-X5X: 8 spatial streams, three radios (only AP-555 for now… that thing is a monster)
  • AP-X6X: Outdoor AP with 2 Spatial streams
  • AP-X7X: Outdoor AP with 4 spatial streams
  • AP-X8X: Outdoor AP with 60GHz (only AP-387)

The last digit indicates the antenna type. Odd numbers are internal, even numbers are external.

  • AP-XX3: Internal Omni
  • AP-XX4: Connectorized
  • AP-XX5: Internal Omni
  • AP-XX7: Internal Directional
  • AP-XX8: Connectorized and ruggedized,

APs with the H suffix indicate a wallplate mount designed for the hospitality industry. These APs also have a built-in switch. I love these APs.

Naturally, if you want to get the gory details, head on over to Aruba and look for the data sheet.

Stay tuned for the next Hands On post in which I will discuss Aruba Central.

Disclaimer: Aruba is my employer, but this post reflects my personal experience as a wi-fi nerd with Aruba products. Some APs were purchased on the open market, some were provided to me by my employer for lab use. This is not a paid promotion, and is not official Aruba communication. I am not part of the Instant product team.