All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
We operate 720 servers and we are happy to buy more.
Operable4829
Member
I work for IPinfo.io and admittedly, I think we have servers with most of the vendors here in this forum. But it is worth a shot anyway.
🔗 https://docs.google.com/forms/d/1SS5JIMG97MvMTLRmDlscerajmVN8vscQ1XC5kyEozPg
We use these servers to generate IP location data by using ping triangulation. Unlike every other provider in our industry, we use active measurements and evidence-based data. When it comes to IP data accuracy, we are pretty much uncontested at this point, but we are focused on continuously better. Buying new servers helps us get better.
Because we do ping triangulation and active measurements the big things for us:
- Location diversity
- Network peering
- IPv6
Even if you do not checkmark every criterion, fill up form anyway.
Also, if you have issues with our data for your ranges. Let me know as well. Happy to help 
Comments
Did you have server also with @calin?
@Hotmarer I am not sure. The form is pretty small. The vendor can submit it in under 5 minutes. I do reach out to every vendor with an email to let them know what decision we made.
Got one submission already in less than 10 minutes! Thank you!
But $$$ is above data accuracy, huh (crazy you send a correction for those ranges, and they revert once you check their geofeed again
)
Nord has servers in The Bahamas!
https://ipinfo.io/45.95.160.2
Or Belize!
https://ipinfo.io/45.95.162.2
Your "ping" checks are completely useless.
Could you tell, based on your system, where this server is at?
https://ipinfo.io/193.233.233.233
https://ipinfo.io/23.185.104.88 servers in 🇰🇵!
U sure this server is in Frankfurt and just blindly trust that provider's geofeed?
Or this one in Dublin?
Woah! Cloudflare expanded!
What's your criteria for marking an ASN as "Hosting", because as far as I know, this is my ISP.
https://ipinfo.io/AS57269
regards
@zGato I really do appreciate you looking in to our data! We appreciate it.
Please, note the
"anycast": truedata. Anycast IP addresses can not be reliably located in a single location: https://community.ipinfo.io/t/what-is-anycast-how-does-ipinfo-geolocate-anycast-ips/5742This is an interesting case. I think these IPs are located in Florida based on other data points but can not confirm it now 100%. I have opened an internal ticket to investigate. Our location is determined by active measurements, and we have to pick from a number of different data points. We are investigating the issue.
https://community.ipinfo.io/t/why-are-there-geolocation-discrepancies-between-ipinfo-s-data-and-other-providers/350
Not all IP addresses have active measurements available. In those cases, we must fallback to other data points. This is one of those cases.
Admittedly, it may appear questionable, but from our perspective, we cannot simply assign any random location because it is North Korea. In the case of this IP address of a non-pingable IP address, we can at least refer to the geofeed.
The geofeed itself is mentioned in the WHOIS records: https://ipinfo.io/AS36530/23.185.104.0/24#block-whois
The geofeed clearly states:
So, based on this information, we have reported the location. In terms of evidence, we have something to point to and that is the geofeed record.
It is a pingable IP address.
https://ping.sx/ping?t=193.233.233.233
Based on the results at this current time, we think it is in Germany.
If it is not there, please share the location evidence. We will be happy to investigate!
Really appreciate these questions. Please, keep looking into the data and share your insights.
I know how anycast works, I just posted that cause it looks funny you blindly accepted some troll correction.
You do not check those IPs since you just grab the data from a geofeed Nord has sent you. They're the so-called "virtual locations". Those are 100% in Florida, and have always been. You just upstream to Maxmind and sometimes "trust" them, but Maxmind is quite the worst in terms of accuracy, at least for "hosting" IPs.
So you just confirmed you blindly check geofeeds and trust them. Like Maxmind basically, and I'm sure you know how Maxmind is with their "accurate" data.
Never though about anycast?
Download your own database and filter by ",KP,". Just with that we can argue about your accuracy and so.
I must say that outside of Nord and other VPN's cases (which I still think you're getting $$$ from them to report that data), you're, by far, the most accurate database.
We are IPv9 only and payment is 🥭 only.
Do you want our servers?
@zGato Thank you again!
We use a number of different criteria to categorize an ASN. But the simplest thing we can look into is publicly available to you on that page.
Nearly two thousand domains are hosted on your ASN. From our data pipeline point of view, this ASN is categorized as a hosting ASN.
I have opened a ticket for the engineering team to investigate this issue on Monday.
We point to Jakarta, ID. I have not seen this IP address located in North Korea. I am not sure when you took the screenshot, to be honest. But still, it is an anycast IP address, so a singular location should not be considered valid.
We have a list of all the anycast IP address here: https://ipinfo.io/tags/anycast
I literally wrote a community post about this. I investigated virtual servers and was planning to write a series on it.
https://community.ipinfo.io/t/the-lie-in-their-whois-ip-geolocation-reporting-explored/4599
We have no association with MaxMind. We produce our own original data.
Hence the fact why I am here, asking to providers to get new servers so we can keep investing in improving our data accuracy.
Nope. In the case where active measurements are not available, we have to fallback to a data source, which in this case is Geofeed. It is quite difficult for us, and everyone else except for the ASN, to verify the location of the IP address when they leave no clue for us.
Other IP geolocation providers use Geofeed as their primary data source. Our primary data source is the latency data we produce from the 720 servers we operate. So, there is no question about blind trust at all!
We are not going to discard geofeed, and we do respect ASNs that diligently maintain them.
https://community.ipinfo.io/t/appropriate-geofeed-urls/5602
Really interesting. Thank you. That's why I appreciate your insights. I will open another internal ticket for the team. They will investigate the issue.
ha ha I know! There are many IP addresses "located" in Antarctica and North Korea. However, you have to consider the "fallback" value aspect. If there is no way to have active measurement, we have to choose a data point. We cannot pick a random location just because the geofeed says North Korea and place them in the middle of nowhere. That is just bad data at that point. There is a hierarchy of data points. In this case, we have to choose something and geofeed, even though it may be inaccurate, it is still a data point nonetheless.
What do you suggest we can do here? How can we improve the user experience and communicate these issues better?
Even Malian hosting providers accept PayPal.
07/08/2024 (M/D/Y) to be exact.
Your whole https://ipinfo.io/countries/kp page was all messed up
You also said North Korea had routers from South Korea Telecom lol
Your "other providers say this IP is in X" is based on MaxMind data. Never said you're affiliated with them, but use their data for your own stuff too.
So stop grabbing Nord/Surfshark and others because they obviously fake them. Or force measurements to IPs that you get the data from MaxMind, cause MaxMind is dog.
Another example!
traceroute. You already do that for some people, but don't for others
@zGato Thank you again!
I will definitely take a look at our historical data but honestly, this makes sense. Adding location exceptions creates a bias in the data. When it comes to anycast IP addresses, we actually have an array of locations. But we have to present a single location to the user.
MaxMind uses public data like the rest of the industry; we do not even have to refer to their data to prove a point. Instead, we can point to geofeed data or WHOIS data to demonstrate why we are better.
We do have to have the existing capacity to parse geofeed and WHOIS data, but we extend on this functionality with active measurement. We can refer to our own public records based data to depict what the rest of the industry thinks about IP location.
We have to demonstrate that we are different and better than others. If we do not compare ourselves to others in the industry, it is very hard to convince users why we are better.
We also have this page: https://ipinfo.io/accuracy
I know and I understand. That is why we are investigating this issue on Monday, to understand how this could have slipped by. I do not have the data in hand to refute the claim or even to share doubts about our data at this moment.
There was one massive VPN that had 300 virtual servers and they were advertising every
/23in different locations. We said they had legit 3 data center locations and rest of the locations were fake. We are intimately familiar with the issue you have presented and have discussions about it continuously.But please understand the fact that we operate a massive data pipeline with tons of data points. For us, improving our data is a continuous endeavor. And feedback like the one you are sharing is extremely valuable to us. We will look into this.
Thank you, that is a great suggestion. We do have traceroute data feed as well, and it is one of our data points. The philosophy of the IP location reporting is to pick a data point and consequently one location. If it is traceroute-based location gets priority; we ignore the ping-based location. That is why anycast location creates a messy situation as we discussed so far.
For 99.99% of users, they just want a single text string of where their IP address is located. We can present them with a dictionary with data points and weights for location hints. But it is not going to be generally useful.
Again, I really am appreciating the feedback from you. I have opened 3 engineering tickets so far. Our team is looking forward to them already.
An you offer massive discount on your service?
No discount unfortunately
A good chunk of our data is freely accessible through the API, data downloads and the website. If you have random requests I am available to help anytime.
We are a reliable and long term customer for the platforms that we partner up with.
Nice page, but again, you're using MaxMind data, I don't know why you don't just accept it. I have some manual corrections I sent to MaxMind that got accepted and a few days later my IP is on your page with that banner. It's quite obvious.
You're clearly better than them, they don't even check corrections in the first place, but your data upstream is clearly MaxMind in case you don't grab from geofeeds, WHOIS or whatever else. What I want to say is that their database is regularly updated in your system for whatever reason.
I must say I don't understand how you can actually mess up in that page with only 25 tests. A random IP I grab from there you claim is from Sweden is actually in Estonia
I understand, but honestly, there are not many giant VPNs in the industry, you should do better on the biggest ones at least. NordVPN/Surfshark can't just slip by.
I think Nord actually assigns whole /24s for virtual locations cause they are rich, but some others just do multiple, for example, /28 or /26 from a single /24.
And I say again that you're by far the most accurate database, and I'm happy for that, just pointing out some flaws and some inconsistencies between hosts/IPs.
I almost forgor
@OwnTN has servers in Pavlodar, KZ and Montevideo, UY.
@Mahfuz_SS_EHL (which seems affiliated to MetroVPS?) in Dhaka, Bangladesh.
@Operable4829 would be helpful if we could get a bigger version of the location map.
So, rough theory, how does the triangulation works?
Didn't know you had this sick map creation.
Took me a while but added my +300 servers
https://ipinfo.io/tools/map/4971e447-6388-4912-9958-1c0fc307a3e5
If you're interested in any of those locations, send me a DM, and I'll give you the provider name.
I'm also open to DMs from providers in different locations than those
Server in Pavlodar?! How did you obtained that and for how much?
@OwnTN
Yes, MetroVPS has infrastructure aka servers at Dhaka, Bangladesh.
Who is your provider in KG, BY?
No norway servers? Quite surprised you don't have Gigahost (terrahost)
BY: https://adminvps.ru/vps/vps_belarus.php
KG: https://vdska.ru/en/
I missed their last yearly VPS promo cause I was sleeping, so...
I would love a server there but not many cheap providers
@terrahost
Amazing way to collect data and engage with the community while doing it!
Do you find you have trouble with some unique locations where the service is essentially long lined or do you avoid those?
Expensive. But their IPs are ISP types. Anti-fraud systems can't detect proxies. Even rDNS is the same as locals.
I also run a hosting search engine for internal purposes
Here are examples for KG
https://bill.nsp.kg/manimg/showroom/1/index.html
https://hoster.kg/vps-hosting-2/
https://prohost.kg/
Belarus is huge... may drop if need.
Hehe, if I remember correctly I even benchmarked those two (on invitation/request from provider). After seeing the result, I seem to remember that my judgement was "not an attractive product unless some desperately need that location (and are in the region themselves at least)" *g
You can also get from Kazakhtelecom directly in Almaty instead. Residential IP too, impossible to detect, it's literally a random IP from their dynamic pool. Just disable ICMP reply, change common ports (or just don't expose any at all) and that's it. At least IPInfo won't flag your IP if you do this.
Payment-wise, you may find it hard. I had to open a ticket and wait for months. own.tn is just PayPal and Stripe so it's quite easy.
Uruguay has great connectivity to South America, and Spain for example (they have direct cables to here) but most people just go with São Paulo.
Pavlodar IMO is if you want to have that exotic location. Latency-wise, most stuff goes directly though Moscow or some other nearby PoP. Also works to serve nearby countries.
It's quite rare to find EPYC CPUs in Kazakhstan from any host. Uptime and network-wise has been great so far.
Just for comparaison, I have AMD EPYC 7313 from own.tn in Pavlodar and Xeon E5-2620 from Kazakhtelecom in Almaty.
Hey @zGato!
I think it is possible that we are using the same data source, which is public records. We are trying to do everything in our power to be better than everyone else.
That's why we really appreciate your feedback. I think we collect a lot of data and are progressively working on improving. Pointing this out triggers tickets and engineering can look into why we are wrong. It will be extremely easy to push a hotfix correction whenever we get feedback from our users. But in reality, whenever we see issues like this, we trigger a whole team-wide motion to investigate, understand, and solve these issues.
Sorry for being a bit misleading with my statement. That VPN was software is part of a parent company with a market capitalization of several digits long. They operate a portfolio of VPN products. You would assume that they can afford to buy an entire Tier 3 data center where they have advertised locations. But nope, it is just three data centers which are rented, I believe.
Tricky stuff. But we are only scratching the surface. The data we sometimes see are quite strange and do not make intuitive sense.
I genuinely and absolutely appreciate your feedback. The issue with us often is that our product is "too loved," I would say. We genuinely do appreciate our users for it. However, consequently, we often do not have great users like yourself going through our data and pointing out places where we can do better. This type of constructive criticism really does help us a lot and keeps the fire burning, and not be complacent.
That is incredible. We do not have the capacity to extract IP addresses from the maps.
Can you share the IP address of the servers through inbox? You can anonymize the last 3 digits if you want. Aside from the location, we also consider latency, peering, and the significance of the ASN, which is why we need the IP address.
If you like our map, you will surely love the summerize tool , and of course the CLI (which I would be surprised if you are not already using!)
TLDR is that overlapping RTT area from different servers.
This article should give you a good overview: https://ipinfo.io/blog/probe-network-how-we-make-sure-our-data-is-accurate/
Essentially, when we ping an IP address, we get a Round Trip Time (RTT). This gives us a radius value based on the RTT from which we can draw a circle on a map with the server being in the center. Then, when we ping the same IP address from another server, we get a new RTT and consequently draw another circle on the map. By pinging the IP address from multiple servers, we get multiple circles, and the overlapping area between these circles is where we think the IP address is located.
Here is a diagram I have on LinkedIn: https://www.linkedin.com/feed/update/urn:li:activity:7247612230760865794/