A high performance in-memory key-value item storage engine written in golang. The storage engine is based lock-free data structures and scales well with multicore CPUs.
- Operations: insert, delete, iterator (lookup, range queries)
- Supports multiple concurrent readers and writers which scales almost linearly
- Database snapshots which facilitates stable and repeatable scans
- Lock-free data structures ensures that concurrent readers and writers doesn't block each other
- The memory overhead of metadata of an item is 64 bytes
- Fast snapshotting: Minimal overhead snapshots and they can be created frequently (eg,. every 10ms)
- Optional memory manager based on jemalloc to avoid golang garbage collector for higher performance
- Custom key comparator
- Fast backup and restore on disk
// Create a nitro instance with default config
db := nitro.New()
defer db.Close()
// Create a writer
// A writer should be created for every concurrent thread
w := db.NewWriter()
for i := 0; i < 100; i++ {
itm := []byte(fmt.Sprintf("item-%02d", i))
w.Put(itm)
}
// Create an immutable snapshot
snap1, _ := db.NewSnapshot()
for i := 0; i < 100; i++ {
if i%2 == 0 {
itm := []byte(fmt.Sprintf("item-%02d", i))
w.Delete(itm)
}
}
// Create an immutable snapshot
snap2, _ := db.NewSnapshot()
// Create an iterator for a snapshot
it1 := snap1.NewIterator()
count1 := 0
for it1.SeekFirst(); it1.Valid(); it1.Next() {
fmt.Println("snap-1", string(it1.Get()))
count1++
}
// Close snapshot and iterator once you have finished using them
it1.Close()
snap1.Close()
// Create an iterator for a snapshot
it2 := snap2.NewIterator()
count2 := 0
for it2.SeekFirst(); it2.Valid(); it2.Next() {
fmt.Println("snap-2", string(it2.Get()))
count2++
}
// Close snapshot and iterator once you have finished using them
it2.Close()
snap2.Close()
fmt.Println(count2 == count1/2)
Apache 2.0