Mastering Ehcache: A Deep Dive into Java’s Lightweight Local Cache Framework

Core Components

Ehcache consists of a Cache Manager that can host multiple cache instances, each implementing the Ehcache interface. A Cache holds the actual cached data, while an Element represents a single cache entry. The underlying storage is accessed through a System of Record (SOR) , which may be business logic, external service calls, or a database.

Data Flow

Flush : Moves cache entries from higher layers to lower layers (e.g., memory to disk).

Fault : Copies an object from a lower layer back to a higher layer when a cache miss occurs.

Eviction : Removes entries from the cache according to the eviction policy.

Expiration : Marks entries as expired based on time‑to‑idle or time‑to‑live settings.

Pinning : Forces an entry to stay in a specific cache tier.

Configuration

Ehcache is configured via an XML file. Below is a minimal example that defines a disk store, default cache settings, and two named caches with different policies.

<ehcache>
  <!-- Directory for disk persistence when data is written to disk -->
  <diskStore path="java.io.tmpdir"/>

  <!-- Default cache configuration applied to all caches unless overridden -->
  <defaultCache maxElementsInMemory="10000" eternal="false" overflowToDisk="true"
                timeToIdleSeconds="0" timeToLiveSeconds="0"
                diskPersistent="false" diskExpiryThreadIntervalSeconds="120"/>

  <!-- Specific cache definitions -->
  <cache name="CACHE1" maxElementsInMemory="1000" eternal="true" overflowToDisk="true"/>
  
  <cache name="CACHE2" maxElementsInMemory="1000" eternal="false"
         timeToIdleSeconds="200" timeToLiveSeconds="4000" overflowToDisk="true"/>
</ehcache>

Key attributes include maxElementsInMemory, eternal, overflowToDisk, timeToIdleSeconds, timeToLiveSeconds, and memoryStoreEvictionPolicy (LRU, FIFO, LFU). Disk persistence requires cached objects to be Serializable.

Features

Performance : Multi‑threaded design optimized for high‑concurrency scenarios.

Simplicity : Small JAR, zero‑dependency configuration, ready‑to‑use out of the box.

Multiple Caching Strategies : Supports various eviction policies and expiration settings.

Two‑Level Storage : Combines in‑memory and optional disk storage, allowing larger data sets than pure memory caches.

Monitoring : Provides listener interfaces for cache and cache‑manager events.

Scalability : Allows multiple cache‑manager instances and multiple cache regions per manager.

Timeout Management : Timeout settings apply to the whole cache; fine‑grained per‑key expiration is not provided, and expired entries may linger in memory, increasing leak risk.

While Ehcache supports disk persistence, data loss can still occur if the in‑memory tier is not explicitly flushed (e.g., via cache.flush()) before an application crash, because only serialized objects can be written to disk.