VictoriaMetrics/lib/mergeset/table_search_timing_test.go
Zakhar Bessarab 837d0d136d
lib/mergeset: add sparse indexdb cache (#7269)
Related issue:
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7182

- add a separate index cache for searches which might read through large
amounts of random entries. Primary use-case for this is retention and
downsampling filters, when applying filters background merge needs to
fetch large amount of random entries which pollutes an index cache.
Using different caches allows to reduce effect on memory usage and cache
efficiency of the main cache while still having high cache hit rate. A
separate cache size is 5% of allowed memory.

- reduce size of indexdb/dataBlocks cache in order to free memory for
new sparse cache. Reduced size by 5% and moved this to a separate cache.

- add a separate metricName search which does not cache metric names -
this is needed in order to allow disabling metric name caching when
applying downsampling/retention filters. Applying filters during
background merge accesses random entries, this fills up cache and does
not provide an actual improvement due to random access nature.


Merge performance and memory usage stats before and after the change:

- before

![image](https://github.com/user-attachments/assets/485fffbb-c225-47ae-b5c5-bc8a7c57b36e)


- after

![image](https://github.com/user-attachments/assets/f4ba3440-7c1c-4ec1-bc54-4d2ab431eef5)

---------

Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
2024-10-24 15:21:17 +02:00

116 lines
3.1 KiB
Go

package mergeset
import (
"bytes"
"fmt"
"math/rand"
"os"
"sync/atomic"
"testing"
)
func BenchmarkTableSearch(b *testing.B) {
for _, itemsCount := range []int{1e5, 1e6, 1e7} {
b.Run(fmt.Sprintf("itemsCount-%d", itemsCount), func(b *testing.B) {
benchmarkTableSearch(b, itemsCount)
})
}
}
func benchmarkTableSearch(b *testing.B, itemsCount int) {
r := rand.New(rand.NewSource(1))
path := fmt.Sprintf("BenchmarkTableSearch-%d", itemsCount)
if err := os.RemoveAll(path); err != nil {
b.Fatalf("cannot remove %q: %s", path, err)
}
defer func() {
_ = os.RemoveAll(path)
}()
tb, items, err := newTestTable(r, path, itemsCount)
if err != nil {
panic(fmt.Errorf("cannot create test table at %q with %d items: %w", path, itemsCount, err))
}
// Force finishing pending merges
tb.MustClose()
var isReadOnly atomic.Bool
tb = MustOpenTable(path, nil, nil, &isReadOnly)
defer tb.MustClose()
keys := make([][]byte, len(items))
for i, item := range items {
keys[i] = []byte(item)
}
b.Run("sequential-keys-exact", func(b *testing.B) {
benchmarkTableSearchKeys(b, tb, keys, 0)
})
b.Run("sequential-keys-without-suffix", func(b *testing.B) {
benchmarkTableSearchKeys(b, tb, keys, 4)
})
randKeys := append([][]byte{}, keys...)
r.Shuffle(len(randKeys), func(i, j int) {
randKeys[i], randKeys[j] = randKeys[j], randKeys[i]
})
b.Run("random-keys-exact", func(b *testing.B) {
benchmarkTableSearchKeys(b, tb, randKeys, 0)
})
b.Run("random-keys-without-suffix", func(b *testing.B) {
benchmarkTableSearchKeys(b, tb, randKeys, 4)
})
}
func benchmarkTableSearchKeys(b *testing.B, tb *Table, keys [][]byte, stripSuffix int) {
for _, rowsToScan := range []int{1, 10, 100, 1000} {
b.Run(fmt.Sprintf("rowsToScan-%d", rowsToScan), func(b *testing.B) {
benchmarkTableSearchKeysExt(b, tb, keys, stripSuffix, rowsToScan)
})
}
}
func benchmarkTableSearchKeysExt(b *testing.B, tb *Table, keys [][]byte, stripSuffix, rowsToScan int) {
searchKeysCount := 1000
if searchKeysCount >= len(keys) {
searchKeysCount = len(keys) - 1
}
b.ResetTimer()
b.ReportAllocs()
b.SetBytes(int64(searchKeysCount * rowsToScan))
b.RunParallel(func(pb *testing.PB) {
r := rand.New(rand.NewSource(1))
var ts TableSearch
ts.Init(tb, false)
defer ts.MustClose()
for pb.Next() {
startIdx := r.Intn(len(keys) - searchKeysCount)
searchKeys := keys[startIdx : startIdx+searchKeysCount]
for i, key := range searchKeys {
searchKey := key
if len(searchKey) < stripSuffix {
searchKey = nil
} else {
searchKey = searchKey[:len(searchKey)-stripSuffix]
}
ts.Seek(searchKey)
if !ts.NextItem() {
panic(fmt.Errorf("BUG: NextItem must return true for searchKeys[%d]=%q; err=%v", i, searchKey, ts.Error()))
}
if !bytes.HasPrefix(ts.Item, searchKey) {
panic(fmt.Errorf("BUG: unexpected item found for searchKey[%d]=%q; got %q; want %q", i, searchKey, ts.Item, key))
}
for j := 1; j < rowsToScan; j++ {
if !ts.NextItem() {
break
}
}
if err := ts.Error(); err != nil {
panic(fmt.Errorf("BUG: unexpected error for searchKeys[%d]=%q: %w", i, searchKey, err))
}
}
}
})
}