mirror of
https://github.com/makayabou/asg-server.git
synced 2026-05-02 17:43:36 +02:00
461 lines
10 KiB
Go
461 lines
10 KiB
Go
//nolint:errcheck
|
|
package cache_test
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"math/rand"
|
|
"strconv"
|
|
"testing"
|
|
"time"
|
|
|
|
"github.com/android-sms-gateway/server/pkg/cache"
|
|
)
|
|
|
|
// BenchmarkMemoryCache_Set measures the performance of Set operations
|
|
func BenchmarkMemoryCache_Set(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
key := "benchmark-key"
|
|
value := "benchmark-value"
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_Get measures the performance of Get operations
|
|
func BenchmarkMemoryCache_Get(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
key := "benchmark-key"
|
|
value := "benchmark-value"
|
|
|
|
// Pre-populate the cache
|
|
cache.Set(ctx, key, value)
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Get(ctx, key)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_SetAndGet measures the performance of Set followed by Get
|
|
func BenchmarkMemoryCache_SetAndGet(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
i := 0
|
|
for pb.Next() {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
i++
|
|
|
|
cache.Set(ctx, key, value)
|
|
cache.Get(ctx, key)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_SetOrFail measures the performance of SetOrFail operations
|
|
func BenchmarkMemoryCache_SetOrFail(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
key := "benchmark-key"
|
|
value := "benchmark-value"
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.SetOrFail(ctx, key, value)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_GetAndDelete measures the performance of GetAndDelete operations
|
|
func BenchmarkMemoryCache_GetAndDelete(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
i := 0
|
|
for pb.Next() {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
i++
|
|
|
|
cache.Set(ctx, key, value)
|
|
cache.GetAndDelete(ctx, key)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_Delete measures the performance of Delete operations
|
|
func BenchmarkMemoryCache_Delete(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
i := 0
|
|
for pb.Next() {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
i++
|
|
|
|
cache.Set(ctx, key, value)
|
|
cache.Delete(ctx, key)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_Cleanup measures the performance of Cleanup operations
|
|
func BenchmarkMemoryCache_Cleanup(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
// Pre-populate cache with many items
|
|
for i := 0; i < 1000; i++ {
|
|
key := "item-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Cleanup(ctx)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_Drain measures the performance of Drain operations
|
|
func BenchmarkMemoryCache_Drain(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
// Pre-populate cache with many items
|
|
for i := 0; i < 1000; i++ {
|
|
key := "item-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Drain(ctx)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_ConcurrentReads measures performance with different numbers of concurrent readers
|
|
func BenchmarkMemoryCache_ConcurrentReads(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
key := "benchmark-key"
|
|
value := "benchmark-value"
|
|
|
|
// Pre-populate the cache
|
|
cache.Set(ctx, key, value)
|
|
|
|
benchmarks := []struct {
|
|
name string
|
|
goroutines int
|
|
}{
|
|
{"1 Reader", 1},
|
|
{"4 Readers", 4},
|
|
{"16 Readers", 16},
|
|
{"64 Readers", 64},
|
|
}
|
|
|
|
for _, bm := range benchmarks {
|
|
b.Run(bm.name, func(b *testing.B) {
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Get(ctx, key)
|
|
}
|
|
})
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_ConcurrentWrites measures performance with different numbers of concurrent writers
|
|
func BenchmarkMemoryCache_ConcurrentWrites(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
benchmarks := []struct {
|
|
name string
|
|
goroutines int
|
|
}{
|
|
{"1 Writer", 1},
|
|
{"4 Writers", 4},
|
|
{"16 Writers", 16},
|
|
{"64 Writers", 64},
|
|
}
|
|
|
|
for _, bm := range benchmarks {
|
|
b.Run(bm.name, func(b *testing.B) {
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
i := 0
|
|
for pb.Next() {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
i++
|
|
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
})
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_MixedWorkload measures performance with mixed read/write operations
|
|
func BenchmarkMemoryCache_MixedWorkload(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
benchmarks := []struct {
|
|
name string
|
|
readRatio float64
|
|
goroutines int
|
|
}{
|
|
{"Read-Heavy 90/10", 0.9, 16},
|
|
{"Balanced 50/50", 0.5, 16},
|
|
{"Write-Heavy 10/90", 0.1, 16},
|
|
}
|
|
|
|
for _, bm := range benchmarks {
|
|
b.Run(bm.name, func(b *testing.B) {
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
r := rand.New(rand.NewSource(time.Now().UnixNano()))
|
|
i := 0
|
|
|
|
for pb.Next() {
|
|
if r.Float64() < bm.readRatio {
|
|
// Read operation
|
|
key := "key-" + strconv.Itoa(i%100) // Reuse keys to simulate working set
|
|
cache.Get(ctx, key)
|
|
} else {
|
|
// Write operation
|
|
key := "key-" + strconv.Itoa(i%100)
|
|
value := "value-" + strconv.Itoa(i)
|
|
i++
|
|
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
}
|
|
})
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_Scaling measures how performance scales with increasing load
|
|
func BenchmarkMemoryCache_Scaling(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
benchmarks := []struct {
|
|
name string
|
|
operationsPerGoroutine int
|
|
goroutines int
|
|
}{
|
|
{"Small Load", 10, 1},
|
|
{"Medium Load", 100, 10},
|
|
{"Large Load", 1000, 100},
|
|
{"Very Large Load", 10000, 1000},
|
|
}
|
|
|
|
for _, bm := range benchmarks {
|
|
b.Run(bm.name, func(b *testing.B) {
|
|
// Pre-populate cache
|
|
for i := 0; i < bm.operationsPerGoroutine*bm.goroutines; i++ {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
localI := 0
|
|
for pb.Next() {
|
|
// Simulate random access
|
|
key := "key-" + strconv.Itoa(localI%(bm.operationsPerGoroutine*bm.goroutines))
|
|
cache.Get(ctx, key)
|
|
localI++
|
|
}
|
|
})
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_TTLOverhead measures the performance impact of TTL operations
|
|
func BenchmarkMemoryCache_TTLOverhead(b *testing.B) {
|
|
c := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
key := "benchmark-key"
|
|
value := "benchmark-value"
|
|
ttl := time.Hour
|
|
|
|
benchmarks := []struct {
|
|
name string
|
|
withTTL bool
|
|
}{
|
|
{"Without TTL", false},
|
|
{"With TTL", true},
|
|
}
|
|
|
|
for _, bm := range benchmarks {
|
|
b.Run(bm.name, func(b *testing.B) {
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
if bm.withTTL {
|
|
c.Set(ctx, key, value, cache.WithTTL(ttl))
|
|
} else {
|
|
c.Set(ctx, key, value)
|
|
}
|
|
}
|
|
})
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_LargeValues measures performance with large values
|
|
func BenchmarkMemoryCache_LargeValues(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
key := "benchmark-key"
|
|
|
|
sizes := []struct {
|
|
name string
|
|
size int
|
|
}{
|
|
{"1KB", 1 * 1024},
|
|
{"10KB", 10 * 1024},
|
|
{"100KB", 100 * 1024},
|
|
{"1MB", 1024 * 1024},
|
|
}
|
|
|
|
for _, size := range sizes {
|
|
b.Run(size.name, func(b *testing.B) {
|
|
value := make([]byte, size.size)
|
|
for i := range value {
|
|
value[i] = byte(i % 256)
|
|
}
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Set(ctx, key, string(value))
|
|
cache.Get(ctx, key)
|
|
}
|
|
})
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_MemoryGrowth measures memory allocation patterns
|
|
func BenchmarkMemoryCache_MemoryGrowth(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
|
|
sizes := []int{100, 1000, 10000, 100000}
|
|
|
|
for _, size := range sizes {
|
|
b.Run(fmt.Sprintf("%d_items", size), func(b *testing.B) {
|
|
b.ResetTimer()
|
|
|
|
for i := 0; i < b.N; i++ {
|
|
// Clear cache
|
|
cache.Drain(ctx)
|
|
|
|
// Add new items
|
|
for j := 0; j < size; j++ {
|
|
key := "key-" + strconv.Itoa(j)
|
|
value := "value-" + strconv.Itoa(j)
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
}
|
|
})
|
|
}
|
|
}
|
|
|
|
// BenchmarkMemoryCache_RandomAccess measures performance with random key access patterns
|
|
func BenchmarkMemoryCache_RandomAccess(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
const numKeys = 1000
|
|
|
|
// Pre-populate cache with many keys
|
|
for i := 0; i < numKeys; i++ {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
r := rand.New(rand.NewSource(time.Now().UnixNano()))
|
|
|
|
for pb.Next() {
|
|
key := "key-" + strconv.Itoa(r.Intn(numKeys))
|
|
cache.Get(ctx, key)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_HotKey measures performance with a frequently accessed key
|
|
func BenchmarkMemoryCache_HotKey(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
hotKey := "hot-key"
|
|
value := "hot-value"
|
|
|
|
// Pre-populate the hot key
|
|
cache.Set(ctx, hotKey, value)
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
for pb.Next() {
|
|
cache.Get(ctx, hotKey)
|
|
}
|
|
})
|
|
}
|
|
|
|
// BenchmarkMemoryCache_ColdKey measures performance with rarely accessed keys
|
|
func BenchmarkMemoryCache_ColdKey(b *testing.B) {
|
|
cache := cache.NewMemory(0)
|
|
ctx := context.Background()
|
|
const numKeys = 10000
|
|
|
|
// Pre-populate cache with many keys
|
|
for i := 0; i < numKeys; i++ {
|
|
key := "key-" + strconv.Itoa(i)
|
|
value := "value-" + strconv.Itoa(i)
|
|
cache.Set(ctx, key, value)
|
|
}
|
|
|
|
b.ResetTimer()
|
|
b.RunParallel(func(pb *testing.PB) {
|
|
r := rand.New(rand.NewSource(time.Now().UnixNano()))
|
|
|
|
for pb.Next() {
|
|
key := "key-" + strconv.Itoa(r.Intn(numKeys))
|
|
cache.Get(ctx, key)
|
|
}
|
|
})
|
|
}
|
