We all know the first stage in Wi-Fi design is to gather the requirements. But what if you can’t? Or you know full well that whatever requirements are outlined, they’ll probably change tomorrow. What if gathering all the potential stakeholders together in order to work out what on earth they actually all want is impossible? What if you have to support pretty much any client that’s been on the market in the last 5-10 years? Welcome to designing Wi-Fi for Higher Education.

BYOD has been a thing for some time now, as people expect to use their personal mobile device on the company Wi-Fi but even among corporates that have embraced this, there’s usually a high level of control that underpins assumptions about the network. Employees are often issued laptops/phones that are managed with group policy or some form of Mobile Device Management (MDM). IT can push out drivers and can usually decide what hardware is supported. For genuine BYOD there’s usually a policy to determine what devices are supported and that list is reasonably well controlled, limited to business need.

In the HE sector that doesn’t apply. Yes we have managed laptops, which are known hardware running a centrally controlled build, but the majority of devices on our network are personal devices belonging to students. We provide services to academic departments, but if they decide they’re going to buy whatever they like and then expect us to make it work…. that’s pretty much what we’re there for.

Then there’s the question of what users are going to be doing… and we don’t know. Users and research groups move around. Recently a team with a habit of running complex, high bandwidth SQL queries over the Wi-Fi moved out of their office where the network met their needs to a different building where it didn’t and the first I knew about it was complaints the Wi-Fi was down (it wasn’t down, it was just busy but more on that another time)

Yes there are some communication and policy problems where improvements could be made, but the key to designing a network well for HE is to be flexible and as far as possible do what you can to make things futureproof.

“Hahahahaha…. Futureproof” I hear you guffaw. Indeed, what this means practically is making sure we have enough wires installed for future expansion. Our spec is for any AP location to have a double data socket, and we put in more locations than we intend to use precisely to allow flexibility. This can be a hard sell when the budget is being squeezed, but it has paid off many times, and is worth fighting for.

Some of the key metrics of UK universities focus on the student experience. We prioritise delivering a good service to study bedrooms – something that has required wholesale redeployment of Wi-Fi to some buildings.

And so, dear reader, you’ll realise that we do have some requirements defined. Experience and network monitoring tells us we have a high proportion of Apple iOS devices on the network – so coverage is based on what we know helps iPhones roam well. We know a lot of those devices are phones so we factor that into the RF design. We know how much bandwidth our users typically use – it’s surprisingly little but we do have to support netflix in the study bedrooms.

To allow for the relatively high density of access points required to deliver the service students expect we use all available 5GHz channels across our campus and we use a four channel plan on 2.4GHz – both bad news according to some people, but it works.

Perhaps the most important aspect of providing Wi-Fi for the weird combination of enterprise, domestic, education and unpredictable research demands that Higher Ed brings is to make sure you say “yes”. The second you tell that professor of robotics they can’t connect to your Wi-Fi, a dozen rogue APs with random networks will pop up. Agile, flexible, on demand network design is hard work but it’s easier than firefighting the wall of interference from that swarm of robots…. or is that a different movie I’m thinking of?

802.11r or BSS Fast-Transition is a way of significantly increasing the speed of roaming in Wi-Fi environments. More specifically it’s an enhancement to the Wi-Fi standard that describes how to avoid having to perform a full WPA2 authentication when roaming to a new AP.

For the vast majority of clients Fast-Transition (FT) is no big deal. If a user is idly browsing the web, or their client is performing some background sync tasks, a roam from one AP to another isn’t something they notice.

Where it matters is when there are voip clients, or any application that doesn’t tolerate packet loss or high latency.

In an enterprise environment it’s common to use 802.1X authentication. This is quite a chatty affair, and it takes a bit of time. The hallowed figure often quoted for voip clients is to keep latency under 150ms. The process of roaming to a new AP and then performing 802.1X can take longer than this.

With FT, the network authentication part of the roam is reduced dramatically. However, it does need to be supported by the client and network.

Like many standards, there were some elements open to interpretation. Some network vendors required a client to support 802.11r in order to connect to the network, others did not. But with our Aruba network it was initially necessary for all clients to support 802.11r before you could switch it on.

This has changed over time and most vendors have moved to a position of allowing clients that do and do not support FT to co-exist on the network.

Client support has been mixed. For example Apple’s iOS does support 802.11r but MacOS does not (at the time of writing) but happily coexists with it.

Windows 10 includes support, but this seems to be dependent on the Wi-Fi chipset and driver. Which brings me to the blunt message of this post.

In an Aruba OS8 environment, with a jumbled mix of clients, it’s not yet possible to enable 802.11r. I’ve just tried it and run into a couple of Windows 10 laptops that were no longer able to connect to the network. The symptoms observed were the EAP transaction with RADIUS timing out. The user experience was, of course, “the Wi-Fi is down”.

As I started to take up the mantle of Wi-Fi human for a university campus, it was mentioned that we had “the Ekahau laptop”. This turned out to be a woefully under powered old netbook with Ekahau Site Survey installed. Nobody knew how to use it. So I learned.

Fast forward a few years and I’m an Ekahau Certified Survey Engineer who’s designed and surveyed a lot of our campus using this tool.

Ekahau Site Survey is, as the name suggests, a survey tool. It’s also a Wi-Fi design tool. I’ve used it extensively for both tasks and it’s probably one tool I’d struggle to do without.

One of the strengths of ESS it’s relative simplicity. At the most basic level you import a floor plan, set a scale and then you’re ready to use this for predictive design work, or a real world survey.

Surveying is a matter of walking around a building, while clicking on your location on the floor plan. There’s a technique to this of course, but it really is just a matter of walking the floors.

To use Ekahau Site Survey as a design tool you’ll need to draw on walls, doors, filing cabinets and other attenuation areas as appropriate. Then you can place APs on the plan and ESS will show you what coverage can be expected.

What should be obvious about both predictive and survey data is the pretty visualisations generated by ESS will show you exactly what you’ve told it. If you put junk data into your predictive model by saying all the walls are drywall with a 3dB attenuation, your design isn’t going to work very well when it turns out you have 10dB brick walls. So it’s important to have some idea of the building construction and, ideally, have taken measurements.

Likewise with the survey side, if you’re inaccurate with your location clicking or walk 100m between clicks and force ESS to average the data over too large an area, you’ll get a result that’s not as useful.

In short, you do need to know how to use the tool – just like anything.

A quick mention of WiFi adapters. ESS works by scanning the selected WiFi channels (all available in your regulatory domain by default) and recording information and received signal strength of the beacons transmitted from access points. It’s necessary to have a compatible WiFi adapter that can be placed in monitor mode. Low cost options are available. If you give ESS two or three adapters it will spread the channel scan across these, allowing data to be gathered more quickly. ESS will also use the built in WiFi of your laptop to ping an IP or perform speed tests against an iperf server.

I started with a single USB interface, which I used to bash on door frames, before upgrading to a quick release (lego glued to the laptop lid) USB hub with three interfaces connected. This made the laptop lid too heavy and it would fall backwards.

To counter these first world problems, but also to allow for other interesting functionality, Ekahau made the Sidekick. It’s a neat box containing two dual band WiFi radios (802.11ac), a very fast spectrum analyser, processing capability and storage, and a built in battery.

For surveying Sidekick isn’t a necessity but it makes life much easier. The data gathered is more complete, the laptop battery lasts longer and the spectrum analyser capability turns ESS into a powerful troubleshooting tool.

Situated across the lake, next to a lane that borders some fields, is the outdoor lab site of an ecology project researching moorland management. Fascinating in itself, the team tending a very strange allotment sized plot of land are recording data and processing e-mails while literally in the field.

The site is over 300m away from the nearest external Wi-Fi AP in that part of campus and despite the distance the 2.4GHz band is surprisingly usable providing you stand in just the right place and hold your laptop aloft. Because it nearly works the initial proposal from users was to try building a DIY antenna out of kitchenware and a high power USB wireless nic of dubious legality.

I recommended against this and instead have been able to setup an Aruba AP-275 linked back to the campus network with a point to point wireless bridge.

The wireless bridge in question is a pair of Ubiquiti Networks Nanobeam AC, part of the company’s Airmax range of products. This is the first time I’ve used any Ubiquiti gear on campus but I’ve long been a fan of what can be done for a really modest outlay using the Ubiquiti equipment.

Ubiquiti gets a bad rap among wireless geeks. There’s good reason for this. It’s pretty cheap and their Unifi managed WiFi offering has long lacked features that would really qualify it to be truly ‘Enterprise’. The Airmax gear is also inexpensive, built to a price and, frankly, it can look a bit flimsy. Next to the Aruba AP-270 series the Nanobeam looks almost comical in its lack of weather sealing. However, I put a pair of a previous generation Nanobridge M5 devices up, somewhere in the wilds of North Yorkshire several years ago, and have never had to touch them since. Wireless ISPs like Beeline Broadband have been using affordable gear from Ubiquiti and Mikrotik for years to bring broadband to areas that otherwise end up with DSL speeds little faster than dialup.

I think one of the reasons this gear gets a bad name is the way it’s sometimes used. Ubiquiti make some high gain antennas and it’s very easy to significantly exceed the power levels permitted in a regulatory domain. I’ve come across badly installed, poorly aimed radios where the country has been set to whichever would let the installer turn the power up to a metaphorical 11 (but probably higher than that). Because the equipment is inexpensive and accessible this is probably not a great surprise. There have also been some firmware shockers too, but again bad practices have left radios running in the wild with critically vulnerable firmware.

The Airmax gear may not be engineered like our Aruba external APs but it’s affordable, functional, can certainly be reliable and I have to say it’s a joy to use with a really nice user interface. Ubiquiti also make a management server available called UNMS. It’s still in beta, but it does a good job of providing a single pane of glass for seeing the network status and managing Airmax radios.

The relatively short link distance (indicated 280metres) means the Nanobeams can achieve 256QAM to provide 150Mbps throughput with a 20MHz channel width. It may be a distance that WISP engineers would laugh at… but it’s been a useful problem solver and the hardware cost under £200.

• 

• 

• 

• 

Yes yes yes… We all know that we ran out of IPv4 addresses long ago and IPv6 has been around for 300 years and is the solution to all our problems.

But, we’re still not using it. So why not?

“IPv6 is haaard”

For people who are used to an IPv4 world, with 32 bit addresses you can remember like 178.79.163.251 and broadcast domains with a few hundred addresses, IPv6 can be a terrifying prospect with 128 bit addresses that look like 2a01:7e00::f03c:91ff:fe92:c52b, umpteen billion addresses per subnet, a whole new reliance on ICMP and… well, it’s just different.

I never fail to be slightly surprised how conservative and resistant to change some folk in IT can be. It’s just human nature of course, to keep things working and try to avoid too much rapid change, but still.

Dabbling in IPv6 can result in some odd behaviour as it will generally be used in preference to IPv4. So if you can resolve a IPv6 address but can’t route to it, things break. There are ways to address this and most web browsers do, but people have been caught out and that led to IT departments disabling IPv6 on all Windows systems, for example.

Truth be told, IPv6 isn’t all that hard, so why haven’t we deployed it on our WiFi?

She doesn’t have the capacity Jim!

The problem on our network is address table capacity. Previously I’ve talked about a problem we encountered of filling the arp table on the core routers. Keen not to be burned twice by the same problem, IPv6 presents new challenges.

Firstly there’s the obvious issue of each address being four times the size and therefore needing more memory. Then there’s the issue of just how many addresses you’re dealing with.

With IPv4 a client on our network requests an address via DHCP, is given one, and that’s the end of it.

With IPv6 clients come up with a link-local address (starting FE80:) which can be used to talk to other devices on the same subnet without any configuration being done at all. Then, using SLAAC (Stateless Address Autoconfiguration), the router advertises it’s address and, by virtue of that, the local subnet. Clients then come up with their own IPv6 address based on their hardware MAC address. But this address, built out of the MAC address of the client, can be used to track individual machines as they move around different networks. To avoid this most operating systems will also come up with a privacy address. This is another IPv6 address that’s valid in the subnet but is not based on the MAC address of the hardware. Because there are so many addresses in an IPv6 subnet, the chance of a conflict is… well, you don’t have to worry about it.

These privacy addresses are what the client uses to talk to the world, but the client will also respond on its SLAAC address. Privacy addresses are also changed periodically. When a privacy address changes many clients hold on to the old one in case there’s any incoming traffic still trying to use it.

Long and short of all this is you have to ensure your network equipment can handle the number of IPv6 addresses required for the number of clients being supported.

In our testing so far we’ve seen an average 2.6 IPv6 addresses per client. We think our routers could just handle that for our regular concurrent client count but with no room for growth it’s asking for trouble.

It’s worth mentioning that whilst we really do want to provide globally routed IPv6 addresses to the WiFi clients, this isn’t something we actually need to do right now, but do expect it will be required in the future. And we do have options available if we had to make this work right now, the easiest being to spread the subnets over a few routers so as to avoid the need to replace the core routers. We could also just buy some hardware with enough capacity to handle just the WiFi traffic routing.

This situation though presents an opportunity to look again at our whole network and maybe simplify some of it, using technologies such as VXLAN that weren’t available on the hardware we were using previously.

Suffice to say our Wi-Fi is ready for IPv6, the firewall rules are built and tested, the radius accounting all works… we just need the rest of the network to catch up.

Some people are massively into home automation, with a motor and remote control fitted everywhere they possibly can be. I’ve largely not really seen the point of it. I live in a small house, the light switch is never far away. I’ve also found it baffling that in order to switch on the light I might need an internet connection.

However, I did buy a internet connected heating control system opting for Hive by British Gas. The system is easy to install and designed to be a direct replacement for many UK domestic installations. The wireless side of the system uses zigbee and it consists of a boiler control, wireless thermostat and hub that connects to your network. Overall it’s worked well in allowing me to remote control the ancient heating system in my house, but it isn’t really very sophisticated.

The Hive thermostat is a bit…. basic. As far as I can tell it’s old school in that it runs the heating until the temperature measured by the thermostat reaches the target and then stops. The problem is the radiators are then hot, so the temperature in the room keeps rising and will significantly overshoot the target. Before the room temp has dropped to below the target it can start to feel a little cool, because we’ve just acclimatized to that higher temperature. The result is you turn up the heating and end up running the system at a higher temperature than is necessary to be comfortable.

This is how heating control systems have worked for years, but there’s a much better way. Secure (Horstmann) and a few others implement something called Time Proportional Integral (TPI). I don’t pretend to know how this works, but the result is the heating system runs for shorter bursts, switching a predetermined number of times per hour until the temperature is reached and reducing the overshoot that’s common with simple thermostat control.

We have recently got a place in Northern Ireland which we’re using as a base to visit family more frequently. The controls here use TPI but they’re otherwise a standard wired system and I want remote heating control so I can keep an eye on the temperature to make sure it isn’t getting too cold, and also it would be really useful to be able to turn the temperature up before we visit in winter.

Hive is out for the reasons above. Other systems such as Nest that do clever learning of your life patterns are useless in a building that’s not fully occupied all the time. So I’ve gone to rolling my own using Z-Wave controls and Home Assistant with the hass.io distribution on a raspberry pi.

Z-Wave is a really interesting wireless home automation protocol. Like zigbee it employs low bit rate, low energy RF so devices can be battery powered. It is a proprietary protocol though, unlike Zigbee, and whilst I tend to prefer open standards in the real world having a proprietary chipset in every device means it’s easier to get devices from different manufacturers to work together. I have a USB adapter in my raspberry pi so it can act as the Z-Wave controller but a particularly neat feature is devices can be directly paired.

This means I can setup my thermostat to directly control the boiler switch. The state of the two devices is also reported on the Z-Wave network. There’s a really big win with this. If I had a temperature probe in the room and my automation server had to turn on the heating based on a rule what happens if my automation server fails? No heating. Also, adjusting the heating becomes harder. The beauty of using Z-Wave controls, and directly pairing them, is I can have a normal looking thermostat on the wall, and this directly controls the heating. But then I can control the temperature set point of that thermostat remotely using Home Assistant. I can also override the boiler control should the thermostat fail (unlikely) or the batteries run out (more likely).

This gives me the same level of control I have with Hive, but the system isn’t reliant on the hub device being in the middle, I only need my thermostat and my boiler control to make it work. Then you can start getting into the smarter automation functions. Home Assistant can pull in calendar information using caldav and turn that into a switch. So I can automate whether the heating is set above frost protect based on whether there’s a booking in the house booking diary. Which means I don’t have to worry about a visitor using our house turning the heating up and leaving it.

Using Home Assistant with Z-Wave allows for other neat control options such as lighting, blinds, whatever, in order to make an empty home look less empty. But it also allows for those controls to always have a local activation so when a family member who doesn’t own a smartphone wants to use our house, that’s no problem. Essentially I can build up a home automation system to be as simple or sophisticated as I want, but never need to have an internet connection in order to turn on the bathroom light.

But for now, it’s just going to be the heating.

In a lesson of just how important the LAN part of WLAN is, here’s tale of a little problem that hit our network recently… We ran out of space on the ARP table of our core routers.

“What the hell?” I hear you say, “what is this, 2003?” and well you might ask.

Over that last couple of years we’ve been upgrading our campus network, which represents a very large number of switches, from primarily HP Procurve 2600 series edge with 5400zl doing the OSPF routing to primarily the comware range with HPE 5130 edge and 5900 doing the OSPF then 5930s at the core to replace the previous 5900 (setup in a terrible design that meant we could never upgrade them without bringing down everything). It’s a slow process because we try not to break things as we migrate them and, as I mentioned, it’s a lot of switches.

Our network isn’t really all that complicated, there’s just a lot of edge. So an IRF pair of 5930s as the core router in each datacentre seemed to be just fine.

Very recently we upgraded our Aruba mobility controllers and moved them off the 5400 switches that have been handling all our Wi-Fi traffic and on to some of the new kit in our datacentres.

So far so good.

We then slowly moved the subnets from the 5400s to our 5930s. This work was well over 50% complete when one morning, just near the start of the university term, we seemed to have a problem.

Some Wi-Fi users seemed to have no connectivity. We quickly established that plenty of traffic was flowing from the Wi-FI controllers and through our off site link. Whilst the problem seemed to be fairly serious, it wasn’t affecting hundreds of people, as far as we could tell.

Theories flew round the office as we tried to understand what was happening and why some users seemed to have a perfectly good Wi-Fi connection and layer2 traffic was passing, but they couldn’t do anything useful…. There seemed to be no pattern, and a broken user would suddenly start working with nothing having changed.

It was spotted that we had 16,384 entries on the arp table of our 5930s and this was initially dismissed as a small number, but one of my brilliant colleagues pointed out that it was a rather neat, round number and that wasn’t likely to be a good thing.

It turns out that all the comware switches we’re using as routers, 5510, 5900 & 5930 have a max arp table size of 16,384.

As this term has kicked off we’ve seen higher numbers on our Wi-Fi alone, and the 5930s are also routing for all the servers in our datacentres.

This was a pretty basic problem. We’d just filled up the tables and our routing switches were no longer doing the business.

This issue caught us out because, as previously mentioned, this is quite a small number. The generally highly specced 5930 will handle 288K mac addresses and a route table of over 100K. More significantly the decade old switches we were replacing didn’t have this arp table limitation.

Another reason this slipped past is the ARP table size doesn’t appear on the spec sheets of many switches.

We just assumed these very capable datacentre switches had the horsepower and memory allocation to do what we needed, and assuming made an ass of whoever.

Cisco fans will tell you they’ve long had the ability to allocate the finite amount of memory in a layer3 switch to the tables you need, balancing the finite resources between L3 routes, MAC addresses and ARP tables. Fortunately this functionality is now being made available to the 5930.

In our case this means we can reduce the routing table size (I don’t think we need 10K, never mind 100K+) and give more room for arp entries. We can then try again to move the Wi-Fi subnets and, hopefully avoid problems.

The lesson from this has to be the importance of understanding the spec of a box you’re putting at the centre of a network. I can comment on the importance of assessing the implications of a change, such as moving the routing for 16K clients, but to be honest we’d still have assumed the big switches could handle this.

So, roll on the disruption of multiple reboots to bring our 5930s up to the software version that will do what we need, and in the meantime the venerable 5400zl continues to just work.

Finally, I should stress that our network is really very reliable. Serious outages used to be relatively common place and I can’t remember when we last experienced an unexpected widespread outage. This, again, is why we were caught out by this. The 5930s have just been rock solid and they spoiled us.

I’m sure someone once said you shouldn’t use commas in a headline or title but they’re not the boss of me and I don’t care. So, to business…. Here’s one of the rare posts that isn’t about the WiFi.

As someone who’s done my fair share of live sound and studio audio work, I’m something of a gear head. One of the big developments in the area of digital audio has been networked audio and by far the biggest, and some would say only truly relevant player, is Australian company Audinate with their product Dante.

For those that don’t know, Dante is a networked audio system that allows multiple unicast or multicast streams of high quality, uncompressed, sample-accurate audio to be routed between devices across a network. It’s based around custom chips that are designed and sold by Audinate then used by pretty much every serious audio pro audio equipment manufacturer in the world.

Perhaps the best thing about Dante is that it’s just another network device. It doesn’t need anything special to work, and it will work on standard network switches. It’s perhaps no surprise that Dante has become really popular in AV installs. You can even get POE powered speakers, so rolling out a PA system through conference system corridors can all be done just with a network cable. It’s neat stuff.

But this post is less about Dante itself, and how awesome it is, but more how it can impact design decisions in modern networks.

For this example I’m going to use a large lecture hall – a flexible space that can seat 1500 people and is used for conferences, lectures, public events, etc.

A key consideration of a digital audio system is latency – generally referring to how long it takes audio to make it’s way through a system from, say, a microphone input to the speaker output. The old analogue audio system had effectively no latency. Electrons being inconvenienced along the various mic cables, through the mixer, amplifiers and out to the speakers all happens extremely quickly. In digital systems we have analogue to digital conversion, which takes a bit of time and we  invariably introduce buffers, which hold things up a bit and leads to data waiting around. Almost every part of a digital audio system is slower than analogue. Add all these delays together and the system latency can easily get too long to be acceptable.

The same can be said about networking. Every network device your frame or packet passes through will add a bit of latency.

Dante moves audio across an Ethernet network, and switched Ethernet is really very fast (worth noting that in a standard Dante network doesn’t route across subnets). Dante conceptually supports incredibly low latency settings, but most of the chipsets have a minimum latency setting of 1ms. This means anything receiving a stream from that device will allocate a 1ms buffer and will therefore handle network latency of up to 1ms without any disruption to the audio. If the latency exceeds 1ms then you’re in trouble.

A brief blip might be barely noticeable, but sustained high latency will result in late packets being dropped and particularly unpleasant sounding audio stream.

Having discussed latency, and network switches, let’s get to the core of the issue…

Recently installed Dante equipment wasn’t working very well in the lecture hall. The network switches are HPE 5130 with IRF stacking. There are three switch stacks in the building, connected back to an IRF pair of HPE 5900 switches (acting as a router) with 2 x 10Gb links from each stack in a LACP. Stack1 has two switches, stack 2 has five and stack3 has nine switches.

From an administrative perspective there are three switches in the building, though of course we have to remember there’s really far more… why? because latency, that’s why.

Our problem dante devices are situated in mobile racks, each containing their own switch – this allows them to be easily unplugged and moved around.

It turns out one mobile rack was connected to stack 2 and the other was into stack 3. With 1ms latency configured Audinate reckon Dante is good for 10 gigabit switches or about three 100Mb switches (because Fast Ethernet has higher latency) but we were having problems, so just how many switches was our traffic passing through?

The trouble is, I don’t know. Because our Comware switches are in IRF stacks how traffic moves around between members is inside the black box and not available for analysis. Our stacks are a ring topology and we have two uplinks, I don’t know which way around the ring the traffic is going or which of the two uplinks my it’s using.

So I have to assume a worst case scenario, even if this shouldn’t actually happen. Here goes: My first mobile rack switch is connected to stack2 member 1 and that stack hashes my traffic to the second uplink which is on member 9. For some reason rather than taking the shortest route my traffic makes it’s way the long way around the ring, therefore passing through a total of five switches.

It then heads to the routing switch, also an IRF pair, so let’s assume we pass through both of them and on towards stack3. Again we have to assume the worse case scenario, that the traffic passes through all nine switches.

The end result is that it’s possible the traffic could traverse 18 switches.

It’s hard to put some figures on this. According to the HPE 5130 datasheet the gigabit latency is <5us. What I don’t know is whether that changes as more complex configs are applied. 10Gb latency is an even lower <3us. However we don’t know how and even if the IRF affects this. These switches are blisteringly, mind bendingly fast. Even a much cheaper switch is really, really quick, and it’s why we can build switched networks capable of incredible performance. All hail the inventor of the network switch, who’s name I can’t be bothered to look up.

Even though that blisteringly fast performance x 18 still adds up, but not to much. If our Dante traffic follows the worst case scenario, however unlikely, we still only get to something like 200us of latency. This should be absolutely fine, but it wasn’t reliable. I suspect a big part of this is general network conditions. The more physical switches your Dante traffic passes through, the higher the chances it could face moments of congestion.

Back to reality then. Increasing the latency setting of the Dante device would remedy this at the cost of increasing the total audio system latency. If that’s acceptable, it’s the best choice because it provides a healthy margin. The other alternative is to change the network patching so we limit the number of switches. In this case we’ve been able to keep the critical devices connected to stack3 and the very worst case scenario is we’re down to 11 switches and not touching the uplinks but in fact most Dante devices are linked to the same switch stack member.

At the end of this long, confusing mess of writing then I’m left uncertain quite what was causing the issue. Our theoretical latency was well below the minimum required and yet we ran into issues. Probably the most important takeaway for a network engineer is that when working with something as latency sensitive and uncompromising as Dante, it’s important to consider the number of devices in the network path… you know, just like the deployment guide says.