Ethernet at NANOG 92 | Hacker News Reader

Ethernet at NANOG 92

potaroo.net

71 pointstatersolid20 comments

It's taken a while for spraying and trimming to go from being a crazy research idea to mainstream in UEC, but if you're interested, the old animations from my Sigcomm 2017 talk still give the general idea: https://www.youtube.com/watch?v=BO0QhaxBRr0

In addition to what is in the video, UEC introduces a new libfabric-based transport layer and a new congestion control scheme. Plus a bunch of stuff at the lower layers like link-layer retransmit. Hopefully the specs will reach 1.0 around the the end of the year - unfortunately at the moment you can only get them if you're a member of UEC.

That name NDP looks little bit confusing. For example, wikipedia article https://en.wikipedia.org/wiki/Neighbor_Discovery_Protocol shows NDP as Neighbor Discovery Protocol, even when RFC-s themselves don't use that abbreviation.

Yes, it was a poor choice of name. It was originally just a placeholder in the htsim code, from New Datacenter Protocol, but we avoided choosing a better name for too long and had already talked about it as NDP to enough people (especially research funding people) that we decided it would be confusing to change. With hindsight, we should have changed it earlier. Later on we did change the name of its derivative to EQDS.

Ethernet out here quietly leveling up to 800GbE like it’s no big deal… meanwhile I’m just trying to get my WiFi to reach the kitchen

More. Wires.

If you have coax, MoCA is a stellar option.

I put MoCA and a TP-LINK Omada setup (unified control of all AP's - you define the network, adopt the AP's as they are plugged into the network, and they offer the SSIDs and options you want) in my in-laws' house a couple of years ago. They went from almost-undetectable WiFi in most places to a solid signal throughout the entire house. They love it. They even asked why their ISP didn't do it. I told them that $400 of gear and two hours of expertise in setting it up was not in Charter's budget. My MIL even said oh, we'll pay for the gear, you didn't need to spend that much. No, you won't, it's a gift from me (because it means that my wife and I won't have to worry about nonexistent WiFi when we're at your house).

https://www.amazon.com/s?k=flat+ethernet+cable

I have used a few. But they are just as ugly as a round one. In a bachelor pad, fine. Not going to fly in a house.

I’m talking about houses where running ethernet in the walls is impractical or too expensive.

WiFi 7 will fix that for sure, it says on the box that it ‘will finally make true multi-gigabit WiFi throughout the home a reality with faster speeds, less interference, and better performance for today's many high-bandwidth online activities.”

I won’t hold my breath.

https://www.netgear.com/nz/home/discover/wifi7/

Key problem with all the wifi standards is the bandwidth upgrades are real, but they mostly give you more headroom for dealing with interference.

Which is good, but it's why we always have this delta. It does feel like at some point we might end up with apartment building standards requiring 2.4ghz/5ghz mesh in the walls to attenuate cross-talk signals between apartments or something (and then maybe phone-tower micro antennas to provide signal for 5G).

>It does feel like at some point we might end up with apartment building standards requiring 2.4ghz/5ghz mesh in the walls to attenuate cross-talk signals between apartments or something (and then maybe phone-tower micro antennas to provide signal for 5G).

I've had it both ways, and I can't say that it is better either way.

In dense neighborhoods (especially stick-framed apartments), Wifi is problematic due to co-channel interference, as you've addressed. This can be resolved somewhat by increasing the density of my own access points.

But things don't get completely rosy with isolation, either. My current dwelling has aluminum siding, low E windows, and even some metal lath in the walls. This all conspire to make it resemble a Faraday cage.

Wifi works great indoors, but it's a struggle to even load up the most basic of web pages (like HN) using wifi on my front porch.

It is fixable by adding some low-power access points outside and I probably will do that at some point. It's just still not exactly ideal, either.

Both situations have challenges.

Can't we fix this with better beamforming? Lots of stuff doesn't use it at all, there's probably a lot of spectrum to be gained just with very basic stuff.

If they improve it enough and make it cheaper, the WiFi people could rename anything without beamforming to wifi lite or something so everyone knows to avoid it if possible.

With decent planning though and PoE it's substantially easier to add new APs to cover dead spots due to signal attenuation then it is to try and reduce signal attenuation after everything is built though.

I can fish new wires through walls, I can't easily wrap my building in mesh.

nanog is such a treat. makes me miss the old days on irc. Where do nanog people idle now?

Their mailing list also saved my ass from the Crowdstrike debacle

Agreed, the mailing list is the place to lurk.

some in private Signal groups, for engineer to engineer ISP operational discussion which bypasses usual NOC processes

I think a decent number on here.

#ix

NANOG presentations are posted to:

* https://www.youtube.com/@TeamNANOG

NANOG 92 playlist that will be added to over time:

* https://www.youtube.com/playlist?list=PLO8DR5ZGla8jZuNZqoXor...

Fast:

> The IEEE's P802.3dj work is to define a standard specification for a base capacity of a 200Gbps Ethernet lane, and define the combining of 2, 4, and 8 such lanes, to provide 400GbE, 800GbE and 1.6TbE Ethernet interfaces.