15.12.7 How Does the Internet Continue to Work When Parts of It Fail?
- Was the Internet Designed to Survive Failures?
- What Does "Distributed" Mean?
- Why Are There Usually Several Possible Routes?
- How Do Routers Discover Better Routes?
- Does Every Router Know the Entire Internet?
- What Happens If a Fibre-Optic Cable Is Damaged?
- Can Packets Take Different Routes?
- What Happens If a Data Centre Fails?
- Why Doesn't the Internet Collapse During Natural Disasters?
- Has the Internet Ever Completely Failed?
- Is Reliability Becoming More Important?
- What Can Users Learn from This?
- What Should You Remember?
Short Answer
One of the greatest strengths of the Internet is that it has no single point of control or single path through which information must travel. Instead, it consists of millions of interconnected networks linked together by thousands of routers. If a communication link, router, or even an entire network fails, routers can usually discover alternative paths and continue forwarding packets. This distributed architecture makes the Internet remarkably resilient and allows it to continue operating despite equipment failures, natural disasters, and accidental damage.
Was the Internet Designed to Survive Failures?
Yes.
One of the original design objectives of the Internet was to create a communication system that could continue operating even if parts of the network became unavailable.
Rather than relying upon one central computer or one critical communication path, the Internet was designed as a distributed system. Responsibility is shared among many independent networks, each managing its own equipment while cooperating with neighbouring networks.
This philosophy remains one of the Internet's greatest strengths.
What Does "Distributed" Mean?
A distributed network has no single point whose failure would stop the entire system.
Instead, communication is spread across many:
- routers;
- communication links;
- Internet Service Providers (ISPs);
- data centres; and
- independent networks.
If one part of the Internet becomes unavailable, the remaining parts continue operating.
This is very different from a centralized system, where failure of the central computer would stop the entire network.
Why Are There Usually Several Possible Routes?
Most communication networks are deliberately designed with redundancy.
Rather than providing only one communication path, engineers often install multiple links between important locations. For example, a city may be connected to neighbouring cities by several independent fibre-optic cables. Similarly, major Internet Service Providers maintain numerous interconnections with other providers. If one link becomes unavailable, traffic can often be redirected through another.
Although the alternative route may sometimes be slightly longer, communication usually continues.
How Do Routers Discover Better Routes?
Routers continually exchange routing information with neighbouring routers.
As long as communication links remain operational, routing tables remain relatively stable. If a router detects that a neighbouring link has failed, it updates its routing information. Other routers then learn about the change and calculate new routes where possible.
This process usually occurs automatically without requiring any action by Internet users.
Does Every Router Know the Entire Internet?
No. That would be impractical.
The Internet consists of millions of individual networks. Instead of maintaining complete knowledge of every possible destination, each router stores only the routing information necessary to make its own forwarding decisions. Each router asks a simple question: "Which neighbouring router should receive this packet next?"
By repeating this decision many times, packets eventually reach their destinations even though no single router possesses complete knowledge of the Internet.
What Happens If a Fibre-Optic Cable Is Damaged?
Communication cables occasionally fail.
Possible causes include:
- construction work;
- earthquakes;
- floods;
- storms;
- ship anchors damaging submarine cables; and
- accidental equipment failures.
When this happens, routers normally detect that the affected communication path is no longer available. If an alternative route exists, packets begin travelling along the new path. Users may notice a brief interruption while routing information is updated, but communication is usually restored quickly.
This automatic recovery is one of the defining characteristics of modern internetworking.
Can Packets Take Different Routes?
Yes.
Packets travelling between the same two computers do not always follow identical paths. The route selected depends upon current network conditions. Factors that may influence routing include:
- equipment failures;
- communication delays;
- available capacity;
- network maintenance; and
- routing policies used by Internet providers.
Consequently, two packets sent only moments apart may travel through different cities—or even different countries—before reaching the same destination.
To users, however, this variation is normally invisible.
What Happens If a Data Centre Fails?
Modern Internet services are rarely hosted in a single location.
Instead, many organizations operate several data centres distributed around the world. If one facility becomes unavailable because of equipment failure or a power outage, services can often continue operating from another location. Large cloud computing providers routinely distribute applications and stored information across multiple facilities to improve both reliability and performance.
This geographical distribution complements the resilience provided by Internet routing.
Why Doesn't the Internet Collapse During Natural Disasters?
Although severe natural disasters may damage local communication infrastructure, they rarely affect the entire Internet.
Different communication providers often use separate facilities and independent communication routes. International traffic may also be carried by numerous submarine cables and terrestrial fibre networks. Consequently, damage affecting one region usually leaves many alternative communication paths available.
This diversity greatly improves the resilience of global communications.
Has the Internet Ever Completely Failed?
No.
Parts of the Internet occasionally experience outages. Individual websites may become unavailable. Regional communication networks sometimes fail. Major cloud providers occasionally experience service interruptions.
However, the Internet itself has never experienced a complete worldwide failure. Its distributed architecture makes such an event extremely unlikely.
Instead, problems are normally confined to particular regions, organizations, or services while the remainder of the Internet continues operating.
Is Reliability Becoming More Important?
Absolutely.
Society increasingly depends upon Internet connectivity for:
- banking;
- healthcare;
- education;
- emergency services;
- transportation;
- business operations; and
- government services.
As a result, communication providers continually invest in:
- additional fibre-optic routes;
- backup communication systems;
- improved routing technologies;
- redundant power supplies; and
- geographically distributed computing facilities.
These investments further improve the resilience of the global Internet.
What Can Users Learn from This?
Most people assume that the Internet behaves like a single enormous communication system.
In reality, its reliability comes from exactly the opposite approach. Instead of depending upon one central network, the Internet relies upon countless independent networks cooperating with one another. Every router makes simple local forwarding decisions. Together, those millions of decisions create one of the most reliable communication systems ever developed.
This elegant distributed architecture is one of the principal reasons why the Internet has continued to grow successfully for more than forty years.
What Should You Remember?
The Internet remains operational because it is a distributed network of networks rather than a single centrally controlled communication system.
Routers continually exchange routing information and, when necessary, automatically select alternative communication paths. Combined with redundant communication links, geographically distributed infrastructure, and cloud computing, this enables the Internet to continue operating even when individual routers, communication links, or entire networks become unavailable.
This resilience is one of the defining characteristics of modern internetworking and one of the reasons the Internet has become the world's most dependable large-scale communication system.
Back to reading