DataTerminology.comNEW
Also known as: Eviction Policy
The process and rules for removing cached entries to free space or maintain performance.
DESCRIPTION
Cache eviction defines how and when items are removed from a cache, typically due to capacity limits or freshness policies. Strategies include LRU, LFU, TTL-based and size-based eviction. Proper eviction improves hit ratios and latency while preventing memory exhaustion. Misconfigured policies cause thrashing, stale data, or uneven resource use. Effective designs consider workload patterns, object size distribution, and consistency requirements across distributed caches.
ALTERNATIVE DEFINITIONS
Redis Documentation
When Redis reaches the specified maxmemory limit it can begin to evict keys according to the selected eviction policy.
Describes memory-triggered eviction based on configured policies in Redis.
Google Guava
Caches provide several mechanisms to evict entries: size-based eviction, time-based eviction, and reference-based eviction.
Enumerates common eviction mechanisms implemented in application caches.
RELATED TERMS
Cache Least Recently Used Least Frequently Used TTL Cache Hit Cache Miss Redis Memcached Guava Cache
COMPETING TERMS
RELATIONS
- Broader: Caching
KNOWLEDGE AREAS
Software Engineering Data Engineering Infrastructure Observability & Telemetry
TARGET AUDIENCE
Software Developer Data Engineer DevOps Engineer Solution Architect
EXTERNAL SOURCES
- Wikipedia - Cache (computing)
- Redis - Redis eviction policy overview
- Google Guava - Guava cache eviction mechanisms
TOP QUESTIONS
What is cache eviction?
Cache eviction is the removal of entries from a cache based on policies such as capacity or freshness. It keeps memory usage under control and sustains performance.
Category: Definition
Which cache eviction policies are most common?
LRU, LFU, and TTL-based policies are common. Some systems use size-based or custom heuristics for workloads with large or skewed objects.
Category: Overview
How do I choose a cache eviction policy?
Analyze access patterns, object sizes, and consistency needs. LRU suits temporal locality; LFU fits repeated access; TTL works for predictable freshness.
Category: Requirements
What risks are associated with cache eviction?
Poorly chosen policies can cause thrashing, increased misses, and latency spikes. Evicting hot keys can degrade performance dramatically.
Category: Risks
How does cache eviction differ from cache invalidation?
Eviction removes entries due to resource policies; invalidation removes or marks entries stale due to data correctness or freshness requirements.
Category: Misunderstanding
What is a striking example of cache eviction?
Think of a bookshelf at capacity: removing least recently read books (LRU) makes space for new ones, preserving popular titles.
Category: Example
How is cache eviction monitored?
Track hit/miss ratios, eviction counts, latency, and memory usage. Observability tools and cache logs indicate load patterns and policy effectiveness.
Category: Management
What are common mistakes when configuring cache eviction?
Setting too small capacity, ignoring object size variability, and not tuning TTLs can cause frequent evictions and poor performance.
Category: Mistakes
How does cache eviction impact distributed caches?
Uneven eviction across nodes can cause hotspots. Coordinate policies and shard distribution to balance load and avoid skew.
Category: Opinion