How hard is it to disconnect a country from the Internet, really?
That's the number one question we've received about our analysis of the Egyptian and Syrian Internet blackouts, and it's a reasonable question. If the Internet is so famously resilient, designed to survive wars and calamities, how can it fail so abruptly and completely at the national level?
The key to the Internet's survival is the Internet's decentralization — and it's not uniform across the world. In some countries, international access to data and telecommunications services is heavily regulated. There may be only one or two companies who hold official licenses to carry voice and Internet traffic to and from the outside world, and they are required by law to mediate access for everyone else.
Under those circumstances, it's almost trivial for a government to issue an order that would take down the Internet. Make a few phone calls, or turn off power in a couple of central facilities, and you've (legally) disconnected the domestic Internet from the global Internet. Of course, this level of centralization also makes it much harder for the government to defend the nation's Internet infrastructure against a determined opponent, who knows they can do a lot of damage by hitting just a few targets.
With good reason, most countries have gradually moved towards more diversity in their Internet infrastructure over the last decade. Sometimes that happens all by itself, as a side effect of economic growth and market forces, as many different companies move into the market and compete to provide the cheapest international Internet access to the citizenry.
Even then, though, there's often a government regulator standing by, allowing (or better yet, encouraging) the formation of a diverse web of direct connections to international providers. Here's the problem: increased diversity at the international frontier often spells less money for the national incumbent provider (typically the old telephone company, often owned by the government itself). Without some strong legal prodding and guidance from the telecoms regulator, significant diversification in smaller markets with a strong incumbent can take a long, long time.
Here's a map of the world, with countries colored according to the Internet diversity at the international frontier. We did a census, from our own view of the global Internet routing table, of all the domestic providers in each country who have direct connections (visible in routing) to foreign providers.
As a first cut at a diversity metric, this makes a lot of sense; it's easy to compute, and fairly objective (an NSP either has a foreign transit provider visible in the routing tables, or it doesn't). You can think of this, to first approximation, as the number of phone calls (or legal writs, or infrastructure attacks) that would have to be performed in order to decouple the domestic Internet from the global Internet.
If you have only 1 or 2 companies at your international frontier, we classify your country as being at severe risk of Internet disconnection. Those 61 countries include places like Syria, Tunisia, Turkmenistan, Libya, Ethiopia, Uzbekistan, Myanmar, and Yemen.
If you have fewer than 10 service providers at your international frontier, your country is probably exposed to some significant risk of Internet disconnection. Ten providers also seems to be the threshold below which one finds significant additional risks from infrastructure sharing — there may be a single cable, or a single physical-layer provider who actually owns most of the infrastructure on which the various providers offer their services. In this category, we place 72 countries, including Oman, Benin, Botswana, Rwanda, Pakistan, Kyrgyzstan, Uganda, Armenia, and Iran. Disconnection wouldn't be trivial, but it wouldn't be all that difficult. Egypt falls into this category as well; it took the Mubarak government several days to hunt down and kill the last connections, but in the end, the blackout succeeded.
If you have at least 10 internationally-connected service providers, but no more than 40, your risk of disconnection is fairly low. Given a determined effort, it's plausible that the Internet could be shut down over a period of days or weeks, but it would be hard to implement and even harder to maintain that state of blackout. There are 58 countries in this situation, ranging from Bahrain (at the small end) to Mexico (at the largest end). India, Israel, Ecuador, Chile, Vietnam, and (perhaps surprisingly) China are all in this category.
So is Afghanistan, reminding us that sometimes national Internet diversity is the product of regional fragmentation and severe technical challenges. It's true; the government in Kabul is powerless to turn off the national Internet, because it's built out of diverse service from various satellite providers, as well as Uzbek, Iranian, and Pakistani terrestrial transit.
Finally, if you have more than 40 providers at your frontier, your country is likely to be extremely resistant to Internet disconnection. There are just too many paths into and out of the country, too many independent providers who would have to be coerced or damaged, to make a rapid countrywide shutdown plausible to execute. A government might significantly impair Internet connectivity by shutting down large providers, but there would still be a deep pool of persistent paths to the global Internet. In this category are the big Internet economies: Canada, the USA, the Netherlands, etc., about 32 countries in all.
So, could what happened to Egypt and Syria happen in your country? Hopefully not. But it's an important question that companies ask Renesys about all the time, as they decide which countries might reasonably host their new data centers.
Governments that want to encourage direct foreign investment in ICT should have this in mind as they head to Dubai next week for the World Conference on International Telecommunications. Next to Internet performance and stability, the political risks of Internet disconnection are starting to appear on due diligence checklists, as companies consider where to make their investments in global cloud infrastructure.
Wow, a lot of interest in this subject so far. Thanks for the comments and questions. Keep in mind that we're starting with a simple maturity model of foreign/domestic interconnection diversity — factoring in physical interconnection and the 'hidden entanglements' among the logical relationships is the next step. It's interesting that most people who are suggesting modifications to this model believe that their country is much more vulnerable to disconnection. Nobody has claimed that their country is more resilient than we give them credit for!
I'm collecting a list of most-commented countries, and we'll revisit these in some more detail in a future post.
Also, many have asked for a browsable list of the countries that appear in the map. Here's a table of the ISO country codes for each of the four categories. If you'd like more detail on these countries, the relationships among autonomous system providers in them, the size of the incumbents, their international transit arrangements, and the like, I have to recommend that you check out our Market Intelligence service. If you're trying to understand how the online world really fits together at the provider-customer level, region by region and market by market, Market Intelligence provides the best roadmap of who's connected to whom in the service provider industry, from San Francisco to Sao Paolo, from Benin to Vladivostok.
If you're a registered attendee at the WCIT conference in Dubai this week, send me your accreditation and we'll set you up with a free evaluation account.
|Severe Risk||Significant Risk||Low Risk||Resistant|
|AD AI AN AW AX BB BQ BT CF CI CK CU DJ DM ET FO GD GF GL GM GN GP GY IO JE KM KN KP LS LY MC MF MH ML MM MR NF NR PF PM PW RE SB SO SR SS ST SX SY SZ TC TD TK TL TM TN TO UZ VA WF YE||AE AG AM AZ BF BI BJ BM BN BO BS BW BY BZ CD CG CM CW DO EG FJ FM GA GE GG GI GQ GU HT IM IR JM JO KG KY LA LK LR MA ME MG MN MO MQ MT MU MV MW NA NC NE OM PG PK PY QA RW SA SC SD SL SM SV TG TJ TT UG UY VG VI VU WS||AF AL AO BA BD BG BH CL CN CO CR CY DZ EC EE GH GR GT HN HR HU IL IN IQ IS KE KH KR KW KZ LB LI LT LU LV MD MK MX MZ NG NI NP NZ PA PE PR PS PT RS SI SK TH TR TW TZ VE VN ZA ZM ZW||AR AT AU BE BR CA CH CZ DE DK ES FI FR GB HK ID IE IT JP MY NL NO PH PL RO RU SE SG UA US|