Alright now hook that shit up to the router, don’t forget to create a LAG or you’ll create a broadcast storm, and I’m in a WoW raid in ten minutes so make it fast.
The assumption is that they’re creating a high bandwidth trunk interface to the L3 switch/router, so if they forget to create an aggregate it’ll be two independent interfaces and will down the network (or a port will auto down itself with STP, MSTP, etc. but that’s not as funny)
A router of industrial scale which i see at work has its ports to be l3 ports by default. They don’t down the network as the router rejects config where two ports are given the same subnet… at least the ones i operate at work.
That’s true, the default for layer 3 switches is to have its port set as routable.
The original joke really kinda falls flat with modern tech, but it’s still funny to think about handing a switch to someone with zero knowledge and then watch as they accidentally lock up the environment.
LAG are aggregated interfaces and they can indeed be used to prevent (some) layer 2 loops. LAG as in Link Aggregation Group)
Using 2 non-LAG interfaces between the same 2 devices creates a loop.
In the case of a loop, if you’re running spanning tree, one of these interface will be blocking instead of forwarding, preventing the loop, but also percentile the use of this interface until the topology changes (ie: the current one goes down).
If you’re not running spanning tree for some reason, then both interface will chug along, oblivious to the fact that there’s a loop and broadcast packets will indeed keep being flooded on one and received on the other, again flooded, etc. creating a broadcast storm and impacting performance of the whole layer 2 domain and possibly even crashing devices.
A LAG more or less means the interfaces in the group behave as one big (aggregated) interface.
LAG also means you can push traffic on both interfaces for more bandwidth.
It’s not Santa’s fault, he’s thousands of years old so he probably had his IT stack built out ages ago and never bothered to consider upgrades so he just assumed that 10/100 was still state of the art!
As a developer who knows enough about networking and servers to know when I’m out of my depth, I’m sorry for my colleague. If it’s any consolation we all think they are an idiot as well
Or he could go it operations where every day is “a bad day to stop sniffing glue” because you are the only thing keeping the house of cards up while dev and network squabble over who’s foot cannon broke shit this time.
After setting up my own network, and trying to (kinda sorta) do it the right way (multiple SSIDs, vlan segregation, restrictive firewalls for iot, VPN to a VPS, etc.) — I have so much respect for network engineers. First month with my new router, felt like I “broke the Internet” every other day.
Networking has to be the most confusing and tedious IT work I’ve ever done. I still don’t fully understand all the basics of security. But by far the worst part is that troubleshooting can’t be done like normal programming. Network troubleshooting takes forever, and all you get is a working network. Network work feels so dull even I have a hard time seeing my effort.
No kidding. There’s no debugger. You can’t just set a breakpoint and see what’s going on under the hood. It’s more like playing Russian roulette and hoping you don’t bring the whole network down.
It’s messing with the wiring while it’s still hot and there often isn’t a better way to do it.
Add comment