go-wlru
Thread-safe LRU cache with permanency and context-based expiration
Operational Complexity (Time)
| Operation | Best | Average | Worst |
|---|---|---|---|
| Access | Θ(1) | Θ(1) | O(1) |
| Search | Θ(1) | Θ(1) | O(n) |
| Insertion | Θ(1) | Θ(1) | O(n) |
| Deletion | Θ(1) | Θ(1) | O(n) |
| Snapshot | Θ(n) | Θ(n) | Θ(n) |
Operation Complexity (Space)
| Complexity | Value |
|---|---|
| Best | Ω(2n) |
| Average | Ω(2n) |
| Worst | Ω(n + n log(n)) |
Usage
This is a simple example LRU cache structure made with API request lookup caching in mind. If you decide to use this, do so at your own peril.
Thread Safety
It should be thread-safe on all operations.
Benchmarks
BenchmarkSet100k
Running tool: /usr/local/go/bin/go test -benchmem -run=^$ -bench ^BenchmarkSet100k$ github.com/heucuva/go-wlru
goos: linux
goarch: amd64
pkg: github.com/heucuva/go-wlru
cpu: Intel(R) Core(TM) i7-10710U CPU @ 1.10GHz
BenchmarkSet100k-4 1000000000 0.05225 ns/op 0 B/op 0 allocs/op
PASS
ok github.com/heucuva/go-wlru 0.436s
As long as the data set is relatively small in size (<131072 items), the performance stays fairly good. Once this size is passed, the time to perform access/search/insert/delete operations increase significantly - this is due to the design of the underlying sync.Map implementation, as shown below:
BenchmarkSet1M
Running tool: /usr/local/go/bin/go test -benchmem -run=^$ -bench ^BenchmarkSet1M$ github.com/heucuva/go-wlru
goos: linux
goarch: amd64
pkg: github.com/heucuva/go-wlru
cpu: Intel(R) Core(TM) i7-10710U CPU @ 1.10GHz
BenchmarkSet1M-4 1000000000 0.7867 ns/op 0 B/op 0 allocs/op
PASS
ok github.com/heucuva/go-wlru 42.616s
.svg)