Eviction
Eviction is the removal of stored or cached items according to capacity limits or a retention policy.
Vibe Engineering
Cache eviction is the removal of cached entries according to a policy, such as elapsed inactivity time, to limit retained data.
- How it is used
- Controls retained cache data when lazy loading would otherwise allow memory usage to grow with traffic or dataset size.
- How deep it goes
- Explains percentile selection, production measurement, and miss-ratio tuning, but does not develop eviction internals.
- Examples
- Uses a word-existence cache with five-minute access expiration, statistics recording, request percentiles, and production miss-ratio monitoring.
- What you can do after
- Operate/configure a time-evicting cache and tune its duration using traffic distributions, cache statistics, and memory constraints.
Acing the Frontend Interview
Cache eviction is the removal of cached entries to enforce capacity limits or other retention policies.
- How it is used
- Applies eviction to memory-bound caches that must retain frequently accessed data under capacity limits.
- How deep it goes
- Explains LRU behavior, ordering, and constant-time implementation, while deferring broader policy tuning and stampede handling.
- Examples
- Demonstrates eviction through a TypeScript LRU cache, a React demo, and stale-entry removal during cache operations.
- What you can do after
- Builds from scratch an LRU cache with constant-time access and insertion using a Map and capacity-based deletion.
Amazon Web Services in Action, Third Edition: An in-depth guide to AWS
An eviction is the removal of a nonexpired cache item because the cache has reached its memory limit.
- How it is used
- Uses cache eviction to manage memory pressure in ElastiCache and MemoryDB deployments.
- How deep it goes
- Explains eviction triggers, Redis policies, monitoring thresholds, and capacity remedies without detailing implementation internals.
- Examples
- Uses team leaderboards, 4 KB cached items, a 1 GB cache, and a CloudWatch Evictions alarm.
- What you can do after
- Diagnose cache pressure with CloudWatch and reduce evictions by tuning TTLs, resizing nodes, or adding shards.
Read how Amazon Web Services in Action, Third Edition: An in-depth guide to AWS explains eviction →
What to know first
- cachingCaching is the temporary storage of data or computed results for reuse, reducing the work or delay of retrieving or computing them again.
Related concepts
- percentileconfigured by · A percentile is a value in an ordered dataset or distribution at or below which a specified percentage of observations lies.
- performancekind of · Performance is a measure of how effectively a person, process, model, or system achieves its objectives, including qualities such as output quality, speed, and resource efficiency.
- kv-cache evictionincludes · KV-cache eviction removes selected key-value entries from a decoder's cached attention state to reduce decoding memory usage.
- cachingconfigures · Caching is the temporary storage of data or computed results for reuse, reducing the work or delay of retrieving or computing them again.
Questions about eviction
What is eviction?
Eviction is the removal of stored or cached items according to capacity limits or a retention policy.
Which Manning books explain eviction?
Vibe Engineering, Acing the Frontend Interview and Amazon Web Services in Action, Third Edition: An in-depth guide to AWS.
What should I know before learning eviction?
Manning books that teach eviction usually introduce caching first.
What kinds of eviction are there?
Kinds of eviction include kv-cache eviction and model eviction.
What is eviction a kind of?
Eviction is a kind of performance. Performance is a measure of how effectively a person, process, model, or system achieves its objectives, including qualities such as output quality, speed, and resource efficiency.
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.


