At our school with have hundreds of chromebooks, imacs, windows pc and student phones on it at the same time I thought there’s a limit.
There really isn’t a hard limit, other than the memory in your devices and the number of IP addresses you have (internal, and public). Devices sold as “enterprise” generally have larger tables and can handle many more active devices.
In my last corporate job, we had around 10,000 devices behind one “NAT Pool” gateway.
They spread the traffic from all of those devices across a robust network architecture that is designed to handle lots of traffic from many devices in various locations.
Limit is all relative to what you spec it to handle.
You wouldn’t use a farm tractor to cut the grass in a 0.1 acre HOA house, nor would you use a push-mower to mow 10s of acres of parkland.
Your home network doesn’t do a lot and even most marketed “limits” are mostly made up numbers given you can usually only use 1-2 device per person in the house so even “10 device limit” would be plenty for the stereotypical 2-parents 2-kids media image of a “family” if each was only using 2 devices at once and the rest in sleep mode.
Enterprise networks (which is what a school or medium-to-large business would have) typically will split up the network into smaller ones by type (e.g. guest, regular employee/student personal devices, official enterprise-managed-owned trusted devices, printers, etc) and that helps to split up the load across virtual networks. Then they’ll have many WiFi access points to spread out the load, subnets selected to allow thousands (or 10s of thousands) of IPs addresses in their DHCP pool, network switches capable of handling potentially 40+Gbps traffic at the switch and interconnected by 10+Gbps fiber links and multi-gigabit Internet with a router and firewall (usually separate boxes, not like the all-in-one you have at home) that has a powerful processor to handle the max Internet speed to/from the network and are built to do 1 job but do it very efficiently. They can scale VERY efficiently with proper planning…but replacing the major components will probably also cost more than most people’s cars.
The issue obviously isn’t IP addresses, switches, or any of that, this question comes down to, how can 2000+ devices share wireless bandwidth effectively, because while you can add switch ports and wires all you want, there’s only 3 non overlapping 2.4Ghz channels and a handful of 5ghz.
So the first thing you do is put in a lot of access points, use 20Mhz channels both on 2.4 and 5ghz and make it so adjacent access points never have the same channel…
Then you turn the power down low on the access points so that as devices move around the campus they rapidly disassociate from one access point and roam to the new access point closer.
Then some controller magic as well, the access points tell controllers how full they are and the controller tells the access points to disassociate some devices on the theory that they will roam to less jammed access points.
Then, if you’ve got a really good designer, they put in QoS rules that ensure when bandwidth is scarce the most important kinds of traffic get priority. For example access to the course related websites and not game related websites.
But doing this kind of design is nontrivial and I’ve seen sites set up by professionals that completely suck.
Let me teach you about the magic of subnetting and public vs private IP addresses… (Please don’t make me explain subnetting…)
How can the rail company haul so many people around on a train when your car only fits four people? They have a different type and class of equipment/infrastructure than you do.
They can because they have actual IT people that set up, you know, networks.
I have been managing school networks since 1998 and have studied networks for a few years longer than that. The ELI5 way to answer your question is that for each limitation, there is a work-around. For example, one wireless access point (the “radios” on the classroom ceilings which your chromebooks, phones, etc. talk to) originally should really only have a few devices. But that number became bigger over the years, as new “languages” were invented for wifi. Now we’re around 20-40 devices per access point. We can also put access points into every classroom in order to push the limits further. I remember only being able to support one cart of laptops in a whole wing of a high school 20+ years ago, then a whole cart in every two classrooms 20 years ago, then 20-25 per room 10-15 years ago, etc.
Putting that aside for the moment, I think the limitation you’re most concerned with is that there are a limited number of addresses that Internet Protocol can handle. To handle this, someone invented a new technique to hide a bunch of addresses behind a single address. Think of it like the difference between the mailing address of a one-family house vs. the mailing address of an apartment in a tall building. You address your letter or package to 123 Main St. or to 456 Main Street, Apartment #12.
This is called Network Address Translation, or NAT. You probably have a wifi router in your home that uses NAT. You have a single “on the Internet” address, but everything in your house uses a special set of addresses that are designated as “not on the Internet.” (“Apartment #12” in the previous example.) So you could have 2, 10, or 50 devices like laptops, phones, TVs, doorbells cameras, etc. and they each have an IP address. But then they reach out to the Internet and that means they pass through your router. The router applies NAT and the traffic goes out to the Internet in general with a return address of the router. The NAT system in the router writes a note to itself that says, “when kittenwar.com replies, make sure Joe’s laptop gets the web page they requested, but if Netflix replies, make sure the living room TV gets the video it asked for.” In this way, you can have dozens or even hundreds of thousands of devices all “behind” a single IP address. So even though there are a limited number of IP addresses, there is this big block of IPs that different schools and businesses can (kind of) use at the same time. This is pretty much the only time you can use the same IP address that other people use.
This explanation is extremely simplified. I’m leaving out so much stuff that it’s actually a little inaccurate. But I wanted to get the general idea across. NAT allows different schools and businesses to each use only a single address that is unique to them, but it represents almost all of their devices. Sometimes they also set up a server or two on other addresses because of … well, complicated technical reasons. But most schools put thousands of devices behind a single IP address like that. Then they use other techniques, such as using more wifi access points, to be able to have more devices working inside the buildings at the same time. Using the right combination of techniques will allow scaling up to almost any size. For example, one of my first tasks at a school (in 1998) resulted in a change from a limit of a few hundred computers to several million. Last I heard, they only ever needed a few thousand, but if you’re going to change things, plan for the very long term so you don’t have to do it again!
Let me know if you want a deeper look. I wrote this really quickly and probably missed a few things. I’m happy to help someone who honestly wants to understand.
Each room have an access point, in corridors every 50 feet? all wired back to a switch that can handle it.
Well designed architecture and higher end equipment.
they have multiple access point.
They spread out the workload across multiple WiFi Access Points (AP’s). Each AP is capable of handling a couple hundred devices (depending on manufacturer’s spec). Each WiFi AP is connected via Ethernet to the routing/switching backbone.
With enterprise networking, different tasks are split out into different devices. In home networking you’ll have a router which acts as a router/switch/firewall/AP, where in an enterprise environment, each of those functions is typically its own dedicated hardware device(s) (typically clustered for fault tolerance).
If you are interested in learning about things like this, you should look into the COMP-TIA Network+ certification. It’s a great entry-level program for aspiring network engineers.
gear needs to be sized appropriately to handle the traffic. We have 21 skyscrapers in a hub-spoke converged network setup, all tunneling to 1 egress point in a datacenter, as long as the carriers providing internet to our spoke dmvpn routers is sufficient; we maintain 500+ Mbps internet speed throughout all properties.
They have a few more pieces of equipment than a cable modem and a D-link wifi router from Best Buy.
Unbelievably easily, that’s networking.
What blew me away was sitting in the terminal of a major international airport with many thousands of people running about, so many little business shops, screens everywhere. I can’t imagine how many devices on that massive network and constantly changing with the people just passing through. Yet I got the fastest Internet speeds I’ve ever seen on my phone, over 1Gb/s!
You think that is hard? Try doing a stadium with 66,000+ people.
Either a huge IP pool, or more likely, NAT.
They don’t buy the same equipment that you do for your house.