Why blocking whole countries on the Internet isn't a precise process

Author Christian Reading time 11 minutes

Photo by Yan Krukau: https://www.pexels.com/photo/close-up-of-a-person-holding-uno-cards-9068976/

I just read it again on the Internet. Someone is asking: "Hey, as we do only business in the United States, can't we simply block all other countries and be safer? All our customers and suppliers are located in the US."

This inspired me to write a short post about why this is a dangerous and - let's call it politely - sub-clever idea.

You know what "Internet" means, do you?

The term Internet is short for "interconnected networks". The Internet isn't one big network. It's thousands and thousands of small and bigger networks linked together via so called routing protocols. They transport the information on which routers decide how to route your packet so it arrives at its destination. Routers are, to use an analogy, the traffic signs along the highway. Giving each packet directions on which lane it needs to take to reach its destination. In protocol terms, we speak about iBGP, eBGP, OSPF, RIP v1/v2, IGRP, EIGRP, and so on. The only real distinction is whether these protocols are Intra-AS (routing inside one AS - for example iBGP) or Inter-AS (routing between several AS - for example eBGP) routing protocols.

What is an AS, you ask? AS is short for autonomous system (Wikipedia). That's the technical term for a network under the control of a single entity, like a company. Each AS is identified by its unique number, the ASN. This number is used in routing protocols like BGP to exactly identify to which AS a rule belongs.

And as you must have already guessed by now: None of this respects real-world borders. Packets don't stop at borders. Here in Europe, even we people don't stop at borders. You just have to love Schengen (Wikipedia).

Therefore, the task of only allowing customers from the US is a little bit complicated to set up. Technically spoken. Data packets don't contain information from which country they originated. Just the source IP address.

But.. My firewall-/router-/hosting-/DDoS-/CDN-/whatever-provider provides such an option in the control panel of my/our account? So it must be possible!

I didn't say it couldn't be done under any circumstances. I just said, It's complicated and will constantly cause you pain and money loss.

Even BGP in itself isn't 100% safe and attack vectors like BGP hijacking (Wikipedia) do exist, but due to how BGP works, they are always pretty quickly noticed, and the culprit is easily and clearly identified.

So, when it is possible, how do they do it?

They are taking many and many educated guesses.


Yeah, ok. Sprinkled and garnished with some bureaucratic facts as their starting point for their educated guesses.


Ok, sometimes they outright pay internet service providers or other companies to give them those data. This might or might not be legal under your countries data privacy laws..


Not the answer you expected to read? Yeah, life is disappointing sometimes.

How the bureaucratic layer of the Internet works

Tackle the problem from another angle: Do you know how IP addresses are managed on the bureaucratic layer?

Do terms like IANA, RIR, Ripe and ARIN ring a bell? No? Ok, let me explain.

IANA is the Internet Assigned Numbers Authority. In their words, they "Perform the global coordination of the DNS Root, IP addressing, and other Internet protocol resources" Relevant for us in the article is the "IP addressing" part.

The IANA assigns chunks of IP addresses to the so-called RIRs. That's Regional Internet registry (Wikipedia). Those RIRs are (with their founding dates and current area of operations):

  • 1992 RIPE NCC - Europe, Russia, Middle East, Greenland
  • 1993 APNIC - Asian/Pacific region, Australia, China, India, etc.
  • 1997 ARIN - United States of America, Canada
  • 1999 LACNIC - Mexico, South American Continent
  • 2004 AFRINIC - African continent

These RIRs then provide companies in their assigned areas with IP addresses they can manage themselves. And to make this picture easier I left the ICANN & NRO, two other governing bodies, out of the picture.

As you can see some RIRs were founded later than others. This also means: Even if you filter based on which RIR manages the IP addresses, this isn't set in stone forever. Even if a RIR is responsible for a whole continent this can change.

What these companies, which offer geo-blocking, do is: They look where an IP address is located on the bureaucratic layer. Which RIR is responsible for the IP block? Which companies "own" the IPs? Where are they routed to/announced from. But these are all bureaucratic and technical information. These information can't be mapped 1 to 1 to a country. And these bits of information are extremely volatile.

Side note: And there is no RIR for each single country. The term LIR or Local Internet registry (Wikipedia) does exist. But it commonly refers to your Internet Service Provider (ISP) who assigns your Internet Modem/Router an IP address so you can browse the Internet. This has nothing to do with countries. The Internet itself isn't technically designed with the concept of "countries" or "borders" in mind. Never was and most likely (hopefully!!) never will.

Another problem are the systems who provide these information: Some provide real time information. Others don't. Additionally you don't know which metrics your vendor uses and how the vendor obtains them. And usually they don't make the process how they obtain and classify the information publicly available.

I had customer support agents who, instead of resolving a domain name via the ping or host command typed it into Google and used that information. Sometimes obtaining wrong information which was months old and therefore led to other errors...

And what about multi national businesses?

A company from Germany can have assigned IPs from the ARIN for their US business. Maybe they have a subsidiary company for their US business, but this still makes it a German company. How do you filter that?

Keep in mind: Maybe their US subsidiary was only established for jurisdictional problems and all people working with you are sitting in Germany. Hence mails, phone calls, letters, etc. will all come from Germany.

Additionally this company is free to use the IPs as they like. They can announce their BGP routes as they like. Nothing is preventing them from using IPs assigned by the RIPE NCC in the United States. This is done on a regular level. As especially IPv4 addresses are rare and sometimes IPs need to be moved around to satisfy the ever growing demand.

Side note: The NRO publishes the data of all delegated IP blocks under https://ftp.ripe.net/pub/stats/ripencc/nro-stats/latest/. And the file nro-delegated-stats contains the information which IP blocks were assigned by any RIR. You will find lines that the ARIN (Only responsible for the US & Canada) assigned IPs to an entity in Singapore.

Jan Schaumann used that file to present some cool statistics about IP allocations: https://labs.ripe.net/author/jschauma/whose-cidr-is-it-anyway/

To make the picture more complex: IPs issued by a RIR can be used in any country. Their is no rule nor enforcement that IPs issued by a RIR are only to be used in their sphere of influence. Therefore even that first starting piece of information can differ completely from reality.

Hence my statement that all this geolocation business is based on educated guesses. Yes, many positions will be precise. But the question is "For how long?" and do you really want to make your communication depending on that?

The technical reality

BGP routes themselves can change at any time. There is no "You can change them only once every 30 days." You can change them every 5 minutes if you like. They can even change completely automatically. Heck, they have to change automatically if we want a working Internet. There are always equipment malfunctions..

When I worked at a major German telecommunications provider, we utilized BGP to build an automatic fail-over in case an entire datacenter went offline. Both datacenters announced their routes (how traffic can reach them) via BGP towards the route reflector of our network team. Datacenter A announced with a local-preference of 200, datacenter B with a local-preference of 100. In iBGP the highest local-preference value takes priority. This means: If datacenter A should ever cease to function (the iBGP announcements from that datacenter stop reaching the route reflector) the traffic will immediately go towards datacenter B.

In our case, both datacenters A and B were located in Germany. But that was pure chance. My employer also had datacenters in France, the UK, Spain, etc. and of course also in the US. It just happened that the datacenter where my team was allocated the necessary rack space for our servers were both located in Germany.

So the endpoint can literally change every millisecond. And with it the country where traffic is sent to or originates from.

Of course we did regular fail-over tests. Now think about the following scenario: We are doing a live fail-over test. Datacenter A switched to B and datacenter B happens to be located in France. The traffic will be arriving in France for 5 minutes (the duration of our test). In exactly these 5 minutes a scan from a vendor notices that traffic for all IPs affected by our test will be located in France. The software will write this into its database and happily move along.

How long will that false, inaccurate and outdated information be kept in their database? What trouble will that cause your business?

Looking at it from the other side

Ok, so we clarified why geo-blocking is taking educated guesses with a bit of Voodoo. It is time to look at it from the other side, right? As this is a viewpoint which is regularly forgotten completely.

Let's go with the example above: "Hey, as we do only business in the United States, can't we simply block all other countries and be safer? All our customers and suppliers are located in the US.."

Is this really the reality? Are your suppliers and customers located in the US?

I bet 100% that you haven't even understood why you are making that claim. Most people will look at: "Where do we have stores? Where do we ship? What are our target customers?"

This usually leads to an opinion based on bureaucratic metrics. Or in other words: Delivery and invoice addresses.

But what about the customer in Idaho who just recently moved there from Spain and still uses his/her mail account from a Spanish mail provider?

Have you checked which IPs their email server uses? Are they hosted by a big cloud provider like Google, Azure or AWS? Do you have complete and absolute knowledge on how these biggest three tech companies manage their IPs and hundreds of networks today? Tomorrow? Next week?

Even they don't.

Which measures and workarounds they undertake should a datacenter be down? Or just be in a planned maintenance state?

It's fairly normal that in times of need workarounds are done to ensure customers can use the services, for which they are paying, again as quickly as possible.

Businesses change too

How about your biggest client suddenly stopping buying from you? Are you getting no calls for bids any more?

Could it be that the company you did business with was recently acquired by another company? And now they send all their mail from an entirely different mail server hosted in an entirely different country? Could that be the reason the RFQs (requests for quotations) stopped coming?

How much money will you lose before you notice this error?

Last words

I tried to explain in easy words for non-techie people why geo-blocking is usually bad. Yes, it's used by Netflix and many others. Yes, many products offer some kind of feature to achieve some form of geo-blocking.

But keep in mind: They have to do this for jurisdictional reasons. They bought rights to movies to show in certain countries. The owners of these rights want Netflix to ensure only those customers can watch these movies. Because they themselves sold the exact same rights to at least 25 other companies in other countries. And each of their customers will sue them once they notice that a competitor has the same movie in the same country. Hence, Netflix is trapped in a never-ending cat-and-mouse game with VPN companies that constantly change their endpoints.

I haven't even talked about VPNs. I haven't talked about DNS. I haven't talked about mail. All these require IPs to function. All these add several other layers of complexity. But all these are needed for your business to work in the 21st century.

You won't be more secure by blocking China, Russia, or North Korea from your firewall.

You will be more secure by applying patches on time. Using maintained software products. Separating your production environments from your development/test networks and the networks where the PCs/Laptops of your employees are located. By running regular security audits. By following NIST recommendations regarding password security. By defining a good manageable firewall rule framework. By having a ticket system that makes changes traceable AND reproducible. By introducing ITIL or some ISO stuff if you want to go that route.

Be advised: The bad guys are not just sitting in those countries that you are afraid of. China isn't solely attacking out of China in the cyberspace. No. Probably they utilize a nice hacked internet account from John Doe just around the corner of your shop.

Some links

If you want to read further I can recommend the site https://networklessons.com/. If you want to learn more about BGP you can visit https://networklessons.com/bgp and start from there.