The complete IPv4 address space, mapped | Hacker News Reader

The complete IPv4 address space, mapped

worldip.io

The info, at least for my home ISP, is incorrect. The trust score docks 5 points because I don't have a reverse DNS record on the IP, but I do :-P And the reverse DNS box at the top even correcly reports my rDNS

The site says that the trust score was improved because the allocation for my IP is >10 years old, however my ISP didn't exist 10 years ago. The ASN for my ISP is only 5 years old (they brought some IP blocks because you know IPv4 address exhaustion!)

And very very little data is returned for my ipv6 /48

It seems the site basically doesn't do anything for IPv6 :-(

I see the same problem with rDNS. The box at the top correctly reports the DNS name as "forward-confirmed" but I see a -5 points penalty for missing reverse DNS. Tried with various IP addresses, and I observe the same for many (not all).

Hmm it says my ASN has 512 IPv4 addresses, as the guy who filled out the paperwork for that from ARIN and got a /24 (256 addresses) afterwards, I think this has some major bugs.

Oh and it doesn't even properly consume the geofeed information I have published.

It's got that AI slop visual style - you know, the one you get if you ask Claude to make a website.

I can immediately identify it now. it's painfully obvious

US is pretty inflated because of military and other misc government subnets, would be cool if there was a way to exclude government owned subnets. Excluding cloudflare would also be nice since they force everyone to announce via their proprietary systems rather than the standard BGP protocol and then announcing it under their ASN.

It's also inflated by large companies with dozens of class A networks, but who actually need a total of maybe just a few class C subnets. I once worked for a company with tens of thousands of computers that were using public IP addresses, but they were all completely firewalled, and they used proxies for limited Internet access.

Do you have any examples of large companies that have multiple /8s but can get away with a /21?

Hughes had many class As. Rockwell had a few. McDonnell Douglas had some. Boeing had a few. Boeing eventually acquired all of the above. It's possible that they've divested some of them by now. I haven't worked there for almost 10 years.

I can't find one /8 that is assigned to boeing. I think your idea of IP allocation is very outdated.

Well then they've apparently sold them. They had a lot of /8s 10-15 years ago. I used Boeing as an example of wasted IPv4 space and I guess they've trimmed the waste. If all the other companies that are hoarding IPv4 space did the same, everybody would benefit. Someday IPv6 will be "fully deployed" and the IPv4 space will be a relic from the past.

https://www.hstoday.us/subject-matter-areas/cybersecurity/pe...

e.g.

E.I. du Pont de Nemours (DuPont) 52.0.0.0/8

Ford Motor Company 19.0.0.0/8 Retained by Ford

Prudential Financial 48.0.0.0/8 Retained and used

like, why does Ford need a /8 to make cars? a /20 of public IPs is more than enough for corporate and virtually all of their dealerships

You just listed companies that have one /8 not multiple /8s.

Which company has dozens of class A networks? The only one I'm aware of with two is HP who had 15/8 and 16/8, but I think they returned at least a significant amount of that.

BBN/successors may have held multiple class As at times, but being large ISPs probably used a lot of the space? Various clouds have a lot of space, but afaik, not in the form of whole class As.

Looks like IBM probably had multiple class As through acquisition, but I don't think they still hold them either?

These databases don't seem to be too accurate or in agreement. I've tried to use them To block Russia, china, and a few others which has been a great help in server load but when I do a random check on someone scanning me I still get addresses in Russia and china. There's also the issue of countries leasing or selling blocks of addresses so I'll look at an address that's supposed to be in Africa but the traceroute says Singapore, for example. It's just a mess.

I'm actually the person who created the website and yes, geo location means nothing with the bulk of the IP space.

I'm actually the person who created the website. Of course I used AI. Yes the website is displaying inaccurate information but this was due to my attention moving to collecting and retaining accurate data (a lot harder than most of you'd think). Given the scale of the site and infrastructure behind it there's been a lot of trial and error. https://worldip.io/about if you'd care to read more.

That data collection is about 90% solid right now, once I have it 100% I'll begin updating the frontend to read from the proper tables (currently displaying a lot of stale data). Keep checking back on the site and you'll see those changes.

This is why the website is completely free, no account or pay gates. I do plan to offer as a SaaS in time (whoever commented about batch files, I'll be offering those as well), but again this won't be done until I'm certain the captured data is 100% correct.

Thank you to those of you who gave useful feedback.

I really like the website and respect for supporting IPv6 ;) Are you going to keep scanning the v4 address space to see if meta-data changes? If not the data might go out of date rapidly. I see the site plans to add a paid API. Fair... I'm not too sure if people will buy it when you can probably download text files with much the same data as what's on this site. Though I might be wrong here.

If you wanted to do something genuinely cool you could try to build an algorithmic model of latency between network points on the Internet. Such that it would be possible to estimate latency between any two network paths without sending packets first to measure it. I think I read research somewhere that this is possible and it could have applications in routing, distributed systems, and high performance networking.

Yes, currently 12 POP's. See worldip.io/infrastructure

Some weird data quality issues. Says my /24 is registered and announced in X, but in reality it's only announced in X, and registered in Y. Which would be obvious if you pulled the ARIN whois records.

How accurate is it that the DoD has a handful of class A networks? And WTF do they need them for these days, they aren't actually advertising internal networks on the public internet are they?

An addressing scheme is not the same thing as reachability. And the DOD created the Internet. Why would they force themselves to use private address hacks when there will only ever be up to 256 networks connected?

"640k ought to be enough for anybody"

Your trust score system really should go until you base it on something better than whatever it is based on. Cities and state BGPs and private companies 20+ years old have 50% and 60% scores.

Hey, I have one of those.

IP allocations are inaccurate due to all the VPNs, (CG)NATs and proxies spoofing everything now, the whole internet is tangled up like weeds.

That's a reaction to people trying to use IP address as an identifying signal. The reaction is that it's only an identifying signal for normal users, just like all the rest of the signals. Normal people get blocked by Cloudflare. Sophisticated bots never get blocked.

Love the localhost easteregg!

There's actually 13 Easter eggs :)

Good lord can we please get a rule that projects using AI must disclose having done so? The source code frankly is an abomination.

Reply on news.ycombinator.com