How it works
Fault-tolerant systems continue working when part of the system fails; scalable systems handle growth in users or workload.
Look for redundancy, alternate paths, replication, backups, and load distribution.
Common error
No system is infinitely reliable. Fault tolerance reduces failure impact; it does not remove all possible outages.
Worked example
Find which single-link failure a small network can tolerate.
Data / pseudocode
Links: A-B, B-D, A-C, C-D
Source: A
Destination: DThere are two paths, A-B-D and A-C-D. Losing A-B still leaves A-C-D. Both paths still depend on endpoints A and D, so this topology does not tolerate every possible device failure. Redundancy addresses specific failure modes, not unlimited failure.
Exercises
4.3.1
If B-D fails, give a remaining A-to-D path. If D fails, can either listed path still deliver to D?
Show solution
A-C-D remains after losing B-D. If D itself fails, neither path can deliver to that destination; path redundancy does not duplicate the endpoint.
4.3.2
Explain one benefit and one cost of replicating a service across independent servers.
Show solution
Replication can reduce the impact of a server failure or distribute load. It adds cost and coordination requirements such as consistent data and failover testing.