Unit 6 · The Internet & Computing Systems · Lesson 6.7

Fault Tolerance

A system is fault tolerant if it keeps working when part of it breaks. The internet is fault tolerant for one reason: redundancy. Today you find a network’s weak spots — and price the fix.

Big Idea 4 — Computer Systems & NetworksCED 4.2

Before this lesson

No reading — this page is the reading. Yesterday’s string network is today’s whole lesson.

Where We Are

People→ Bits→ Programs→ Internet→ Security→ Big Data→ Impact

Internet. 6.6 showed traffic rerouting around a cut. Today that property gets its exam name — and its price tag.

The Hook

When does it actually fail? Back to the string network. Your teacher keeps cutting. Somewhere, one cut splits the room in two. Before it happens: which string is it?

The strings that matter most are the ones with no backup.

The Idea

A system is fault tolerant if it can keep working when one or more of its parts fail. The internet is fault tolerant because of redundancy: there are extra paths, so when one fails, routing (6.6) uses another.

The sentence the written response wants: redundancy causes fault tolerance. Extra paths exist → when one link or router fails, packets are routed along another path → the devices can still communicate. Name all three steps.

A single point of failure is a part whose failure alone cuts the network apart. A single path has nothing but single points of failure. A ring has none: break any one link and there’s still a way around.

The tradeoff. Redundancy isn’t free. Every extra link is equipment, installation, and upkeep for capacity you hope you’ll never need. Engineers choose how much fault tolerance a system is worth — a hospital pays for more than a backyard shed.

Try It — Find the Weak Spots

Pick a network. Click links to break them and see if every device can still reach every other. Then predict how many single points of failure it has — and check.

network:

Vocabulary

fault tolerance Internet: The ability of a system to keep working when one or more of its parts fail.
redundancy Internet: Extra components that can take over when others fail. In a network: extra paths. Redundancy is what makes a network fault tolerant.
single point of failure Internet: One part whose failure alone stops the system. Fault-tolerant designs eliminate these.

Check

Five questions. For “how many connections must fail” items, find the device with the fewest connections first.

0 of 5 answered

Impact Check

Fault tolerance beyond networks

Cloud storage keeps copies of your files in several data centers. Airplanes carry backup systems. Your school’s grades are (hopefully) backed up. Every one of these is the same trade: pay for redundancy now, or risk losing everything when one piece fails.

Connections

Came from 6.6 — rerouting is the mechanism; fault tolerance is the property it produces.

Came from 4.7 — tradeoffs again: more reliability costs more, like more accuracy cost more time.

Next in 6.8 — a working path isn’t the same as a fast one. Bandwidth and latency.

Returns in 6.15 — the written response asks how the internet keeps working when a router fails. This lesson is the answer.