x2tyfi

Probably a routing issue. Shot in the dark would be that one of these routers is NATing traffic, and the other router doesn’t have a route to that NAT’d range.

It’s interesting to observe and build LLM-driven solutions in Networking.

The biggest challenges that most of us networking people have are around velocity (how fast we can build and scale networks) and how effectively we can operate them (avoid defects, fix them fast when something breaks).

LLMs are great in both areas. AI helps with deployment challenges by speeding up tooling development and the creation of workflows on orchestration platforms. A manual process step today, say - reserving an IP address in an IP DB — is automated the next day instead of on a backlog for years. This post is an example of that (config-gen/config-deploy).

Operations use-cases are more interesting, IMO, and address the “too many signals” problems that we face. Network substrate telemetry, overlay telemetry, service host metrics, service metrics, customer metrics, recent change data, prior alarms - the list goes on. Being a network operator is not for the faint of heart and is under-mentioned on high stress job lists. AI makes AMAZINGLY good network operations triage agents, since they are able to immediately process so many signals.

Exciting times!

It seems odd that such a tiny proportion would run `discoverable=yes`. Any idea why that is? It makes me question the numbers.

My home lab has grown over the years, now consisting of a physical Proxmox cluster, and a handful of servers (RaspPi and micro hosts). A couple years back I got tired of failures related to host-level Docker issues, so I got a NAS and started using NAS storage for everything I could.

I also re-investigated containerization - weighing Docker Swarm vs K3s - and settled on Docker Swarm.

I’ve hated it ever since. Swarm is a PITA to use and has all kinds of failure modes that are different than regular old Docker Compose.

I’ve considered migrating again - either to Kubernetes, or just back to plain Docker - but haven’t done it. Maybe I should look at Uncloud?

Same, one of my very first gaming experiences. Nothing but great nostalgia!

Huge nostalgia wave seeing some of these. Many of them are hard for me to pinpoint but unmistakably familiar.

My first website was a geocities site back in ~1999, which was dedicated to StarCraft. I wish I could find it now; I remember it being really aesthetically pleasing - it would be funny to see how it held up.

“At the primary moment of disruption, we saw YemenNet (aka TeleYemen) lose transit from BICS (AS6774) and Global Cloud Xchange (AS15412) at 22:32 UTC and revert to satellite service from BusinessCom (AS197206)”

It’s never good when your primary ISP’s only connection out to the world is via satellite

Also curious about this

Amazon Prime’s price hikes have a predictable cadence: * 2014: $79 to $99

* 2018: $99 to $119

* 2022: $119 to $139

We should expect a price hike from $139 to $159 in 2026, assuming the trend continues.

This was not my experience, reflecting on about 10 years of service in AWS network engineering (both as an engineer and manager). I’m at Oracle now, which, by contrast, is orders of magnitude more focused on revenue/spend.

That was their Digits box.

You’re almost always going to bottleneck on your home internet or upstream ISP, rather than this local interface. That being said, you aren’t going to be waiting too long either way, depending on download speed. Deepseek R1 is 671GB. Multiply by 8 to get into bits: 5368Gb At full 10gbps (which, again, you probably won’t get): 5368Gb / 10gbps = 537 seconds to download 537s / 60 = 8.95 minutes. Call it 10m with overhead.

The larger question isn’t if we feel or not. One of the questions is: is our “window” into consciousness occurring before or after decisions are made.

If it’s before, then you can easily tie consciousness and free will together. If not, we are effectively watching videos of our bodies operate. Oh - and there is no spoon.

What was your method of invoking Claude, out of curiosity?

All of these seem valid, too, but they don’t need to be mutually exclusive. I’m all for common sense recommendations - even if it only helps a relatively small percentage of families.

I look at it in a similar light to nutritional guidelines.

It’s surprising that more schools haven’t done this. I suspect that we’ll look back in 10 years with it being common and ask ourselves what took so long.

Because it’s extremely lucrative and strategically valuable to the US

While that may be true, the branding of this particular project seems unbeatable. A literal dumb pipe man with wacky arms. It just works.

Almost always. Most people look at global networks like black magic and don’t realize this.

For most large businesses, 90%+ of major network events are caused by internally-driven network config changes.

Depending on how far along the business is towards automation, a proportion of the 90% can be attributed to a human going “off script” - I.e.: making a change that had not been reviewed.

I love this idea and have considered doing something similar with friends for years. It’s cool to see how far they’ve taken it — much bigger than I had envisioned.

My biggest/only concern - which they gloss over, mostly — is security. Combining networks puts added responsibility on every family that joins. What if friend-X’s kid downloads a virus-riddled torrent, which is capable of multiplying across hosts?

Your own hosts/perimeter can always be protected, but there’s a loss of control with this setup.

This is one of those cases where the work is the fun. It just depends on your personal definition of ‘work’ and ‘fun’.

Why though?

The only use case I can imagine is a legacy game which performs a server search by broadcasting/scanning the local network. And even then - most of the time these games had server browsers.

This is true. But OPs point - while a bit hidden - is that even more rest will result in the temperature lowering, eventually.

And when you achieve that was-lower-then-higher-but-now-lower-again temperature, it should be more juicy

This is a good point, and highlights that there’s an additional variable somewhat out of scope from this test.

1. finish temperature (as mentioned in this article) 2. peak temperature

In these tests, the finish temperature and peak temperature were the same - he sliced while carryover cooking was still climbing. However, when resting, most people achieve a peak temperature (due to carryover cooking), then roll down the other side of the curve to a lowered temperature.

As you imply, a longer rest - in conjunction with protein pulled at just the right time - can likely result with both the perfect (to taste) peak temperature, and a lower (and more juicy) slicing temperature

That’s understandable - the rule of “garbage in, garbage out” certainly still applies. I find that many engineers are capable of gathering the right requirements and content, but struggle with the polish/finish that makes docs more consumable - where LLMs can shine.

Undoubtedly a big opportunity area for LLMs. I’ve recently observed engineers deliver LLM-generated (or iterated) docs that blow away any technical writing they had done in the past.

Network Engineering design docs can be somewhat formulaic structurally, making the LLMs job simpler. I imagine in the near future we’ll just ask them to follow doc templates or reference other designs within a RAG system to ensure there aren’t gaps in the doc, etc.

You don’t need AI for this. You can just write a script to pull the most up to date csv

Couldn’t this logic apply to basically every internal metric across every company?

I’ve had similar excited experiences. On one FAANG offer in 2020, I asked for time to think, as you mention. I already had awareness of pay bands and an idea of the range for this particular role from levels.fyi.

When I came back to the recruiter, I asked how much room she had to negotiate, and asked if she could put me “in the middle of the band”. She immediately said yes and it resulted in a bump of $75k/yr or so.

The thing to remember is that recruiters key metrics are about hires. They desperately want to get people in the door, especially after going through all of the effort to schedule interviews. They will happily help you out within reason.

pontificate on gender-affirming care

Dems have not pontificated on gender-affirming care. It is an insignificant issue that affects a minuscule amount of the electorate. There would be minimal discussion on it if it wasn't for the incessant harping from the right to rile up their base.

It is so simple and effective to weaponize social issues. This is easy to see when you read right-wing discussion: they believe that the left is absolutely obsessed with gender-affirming care, because that is the reality they are fed.

I have a conservative relative who talks about 'wokeness' and gender-affirming care almost non-stop, because he believes that it's being 'shoved down his throat', when in reality, it is right-wing media that is doing the shoving.