gerritjvv / cache-example

Little java caching lib as an example

Simple cache example

This is an example for a simple in memory concurrent cache.

Usage

ExpireCache<String, Integer> cache = new ExpireCache<>();
cache.put("a", 1);

int a = cache.get("a");

// remove "a" from cache
cache.delete("a");

// set a key that will expire in 1 second
cache.put("b", 2, 1, TimeUnit.SECONDS);

int b = cache.get("b");

// cleanout any resources used by cache.
cache.close();

Thread safety

Each instance of ExpireCache is threadsafe.

Enhancements left out

• Monitoring metrics
• Expire on write, on read settings

Performance

The expire cache performance is dependant on its storage.
The storage used here is the Java ConcurrentHasMap, which as a hashmap has a constant lookup time O(1)

The default expire thread does a linear search over all cache instances in O(n).
This means if we have 2 cache instances of each 5K entries, the expire thread will go over all 10K entries one at a time. In practice for a cache where accurate cache expires are not required this should not be an issue.

JMH Benchmarks:

Benchmark                                           Mode     Cnt   Score    Error  Units
BenchmarkTests.benchmarkGet                       sample  117612  ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.00    sample          ≈ 10⁻⁶           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.50    sample          ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.90    sample          ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.95    sample          ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.99    sample          ≈ 10⁻³           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.999   sample           0.007           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p0.9999  sample           0.029           ms/op
BenchmarkTests.benchmarkGet:benchmarkGet·p1.00    sample           0.695           ms/op
BenchmarkTests.benchmarkPut                       sample   91054  ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.00    sample          ≈ 10⁻⁵           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.50    sample          ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.90    sample          ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.95    sample          ≈ 10⁻⁴           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.99    sample          ≈ 10⁻³           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.999   sample           0.013           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p0.9999  sample           0.020           ms/op
BenchmarkTests.benchmarkPut:benchmarkPut·p1.00    sample           0.049           ms/op