Go implementation of the jump consistent hash algorithm[1] by John Lamping and Eric Veach.
[1] http://arxiv.org/pdf/1406.2294v1.pdf
import jump "github.com/lithammer/go-jump-consistent-hash"
func main() {
h := jump.Hash(256, 1024) // h = 520
}Includes a helper function for using a string as key instead of an uint64. This requires a hasher that computes the string into a format accepted by Hash(). Such a hasher that uses CRC-64 (ECMA) is also included for convenience.
h := jump.HashString("127.0.0.1", 8, jump.NewCRC64()) // h = 7In reality though you probably want to use a Hasher so you won't have to repeat the bucket size and which key hasher used. It also uses more convenient types, like int instead of int32.
hasher := jump.New(8, jump.NewCRC64())
h := hasher.Hash("127.0.0.1") // h = 7If you want to use your own algorithm, you must implement the KeyHasher interface, which is a subset of the hash.Hash64 interface available in the standard library.
Here's an example of a custom KeyHasher that uses Google's FarmHash algorithm (the successor of CityHash) to compute the final key.
type FarmHash struct {
buf bytes.Buffer
}
func (f *FarmHash) Write(p []byte) (n int, err error) {
return f.buf.Write(p)
}
func (f *FarmHash) Reset() {
f.buf.Reset()
}
func (f *FarmHash) Sum64() uint64 {
// https://github.com/dgryski/go-farm
return farm.Hash64(f.buf.Bytes())
}
hasher := jump.New(8, &FarmHash{})
h := hasher.Hash("127.0.0.1") // h = 5MIT