Fault tolerance
Fault tolerance is the ability of a system to continue providing correct or acceptable service despite failures in its components or dependencies.
ScyllaDB in Action
Fault tolerance is the ability of a system to continue providing correct or available service despite component failures.
- How it is used
- Frames replicated, distributed storage as the way ScyllaDB remains available when nodes fail.
- How deep it goes
- Explains replication, peer routing, eventual consistency, and acknowledgment thresholds, then verifies quorum behavior by stopping a node.
- Examples
- Uses a three-node cluster with replication factor three, QUORUM queries, and one stopped node.
- What you can do after
- Operate/configure it by choosing replica acknowledgments that keep queries available through planned node failures.
Akka in Action, Second Edition
Fault tolerance is the ability of a system to continue operating and preserve important services when components or operations fail.
- How it is used
- Frames reliable concurrent and distributed applications as recovery design rather than impossible fault prevention.
- How deep it goes
- Explains containment, replacement, reboot, actor supervision, and a clustered node-failure scenario; defers broader catastrophic protection to clustering.
- Examples
- Demonstrates a text-processing cluster whose master accumulates lag while aggregators fail, then resumes delegation after a replacement joins.
- What you can do after
- Apply it via Akka supervision and clustering to keep message-based work progressing through actor or node failures.
Read how Akka in Action, Second Edition explains fault tolerance →
Acing the Frontend Interview
Fault tolerance is the ability of a system to continue providing acceptable service despite component failures or disruptions.
- How it is used
- Frames fault tolerance as an interview-relevant distributed-system goal supported by caching and replicated data.
- How deep it goes
- Explains resilience through caching and replication, including availability, failover, consistency, latency, and operational trade-offs.
- Examples
- Uses multi-tier caches, LRU eviction, primary–replica databases, asynchronous replication, and automatic failover as concrete examples.
- What you can do after
- Evaluate distributed designs by explaining how caches and replicas preserve service, then defend their consistency and complexity trade-offs.
Read how Acing the Frontend Interview explains fault tolerance →
Bootstrapping Microservices, Second Edition: With Docker, Kubernetes, GitHub Actions, and Terraform
Fault tolerance is the ability of a system to continue providing acceptable service when some components fail.
- How it is used
- Builds reliable microservices that continue serving customers during component failures and recover from errors.
- How deep it goes
- Explains replicas, health checks, timeouts, retries, queues, and circuit breakers, with JavaScript and Axios examples but limited internals.
- Examples
- Uses Kubernetes replicas, Axios timeouts and retries, internal services, and a Dropbox integration as concrete failure scenarios.
- What you can do after
- Build it from scratch using replicas, health checks, timeouts, retries, queues, and circuit breakers.
Apache Kafka in Action
Fault tolerance is the ability of a system to continue operating correctly despite failures in components or infrastructure.
- How it is used
- Keeps Kafka event flows recoverable through component failures and guides Kafka Connect worker and error-handling choices.
- How deep it goes
- Explains task reassignment and specific error settings, but does not detail Kafka’s underlying storage failure mechanisms.
- Examples
- Uses a 12-table database import, a worker-crash scenario, faulty connector records, and a retailer splitting workloads across clusters.
- What you can do after
- Operate/configure it by distributing Connect tasks and choosing logging, skipping, dead-letter queues, or retries for connector failures.
Designing Deep Learning Systems: A guide for software engineers
The ability of a system to continue operating correctly or recover when components or operations fail.
- How it is used
- Protects distributed training and hyperparameter-optimization workflows from losing all progress after worker, node, or network failures.
- How deep it goes
- Explains checkpoint storage, worker restart, and state restoration, but does not detail checkpoint formats or recovery orchestration.
- Examples
- Uses remote model-state storage, TensorFlow and PyTorch backup features, and failed sequential HPO trials caused by node or network problems.
- What you can do after
- Operate/configure training services to store checkpoints remotely and resume failed workers or sequential optimization trials.
Read how Designing Deep Learning Systems: A guide for software engineers explains fault tolerance →
Also covered in
- The Coder CafeIn early access (MEAP)
- Server-Side WebAssembly
- Microsoft Azure in Action
- gRPC Microservices in Go
- Grokking Software ArchitectureIn early access (MEAP)
- .NET in Action, Second Edition
- Elasticsearch in Action, Second Edition
- Quantum Programming in Depth
What to know first
- distributed systemA distributed system is a computing system whose components run across multiple networked processes or machines and coordinate through communication.
- availabilityAvailability is the degree or proportion of time that a system or service is operational and accessible when required.
- distributed trainingDistributed training is the coordinated execution of machine-learning model training across multiple computing devices, processes, or machines.
- resilienceResilience is the ability of a system to continue providing acceptable functionality or recover when failures or other disruptions occur.
Related concepts
- resiliencecontrasts with · Resilience is the ability of a system to continue providing acceptable functionality or recover when failures or other disruptions occur.
- stretched clusterproduced by · A stretched cluster is a distributed cluster whose nodes are deployed across multiple physical locations or failure domains.
- transaction logaddressed by · A transaction log is an ordered record of transactions, transaction-related events, or resulting data changes, maintained as a history and often used for tracking or recovery.
- failureaddresses · A failure is an event or condition in which an operation, component, or system does not achieve an expected result or perform its intended function.
- changelogaddressed by · A changelog topic is a Kafka topic that records state-store updates so processing state can be recovered or restored.
Questions about fault tolerance
What is fault tolerance?
Fault tolerance is the ability of a system to continue providing correct or acceptable service despite failures in its components or dependencies.
Which Manning books explain fault tolerance?
ScyllaDB in Action, Akka in Action, Second Edition, Acing the Frontend Interview, Bootstrapping Microservices, Second Edition: With Docker, Kubernetes, GitHub Actions, and Terraform, Apache Kafka in Action and Designing Deep Learning Systems: A guide for software engineers.
What should I know before learning fault tolerance?
Manning books that teach fault tolerance usually introduce distributed system, availability, distributed training and resilience first.
What is the difference between fault tolerance and resilience?
Fault tolerance is the ability of a system to continue providing correct or acceptable service despite failures in its components or dependencies. Resilience is the ability of a system to continue providing acceptable functionality or recover when failures or other disruptions occur.
What is fault tolerance also called?
Fault tolerance is also called fault tolerant.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.





