Documentation ¶
Overview ¶
Package maglevhash implements Maglev Hash algorithm from the paper [Maglev: A Fast and Reliable Software Network Load Balancer](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/44824.pdf) Maglev hash break down the key space to slots and assign slots to nodes. Each node is a logic entity that can handle the key. It could be a server, a service, a VIP, etc. Maglev hash guarantees the following two properties.
- Load Balance. Let N be the number of nodes and M be the number of slots. The difference between any node's assigned slots is less than N/M. For example, if M = 100*N, then the node with most slots has no more than 1% slots than the node with the least slots.
- Minimal disruption. On average, add a new node causes reassigns M/N slots.
Example mh, _ := NewMaglev([]string{"B0", "B1"}) node := mh.Node([]byte("key1"))
Index ¶
Constants ¶
const ( // The default number of slots is the smallest prime number larger than 10,000. // It should be OK for less than 100 nodes. DefaultSlotCnt = 10007 )
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type MaglevHash ¶
type MaglevHash struct {
// contains filtered or unexported fields
}
MaglevHash assigns fixed number of slots to a collection of nodes. 1. Nodes are identified by string. 2. The default hash function hashing key to slot is CRC32.
func NewMaglev ¶
func NewMaglev(nodes []string) (*MaglevHash, error)
NewMaglev creates a Maglev hash for the given nodes, using default number of slots and the CRC32 key hash function.
func NewMaglevWithTableSize ¶
func NewMaglevWithTableSize(slotCnt int, nodes []string, keyHashFn KeyHashFnType) (*MaglevHash, error)
NewMaglevWithTableSize creates a Maglev hash. Nodes are identified by strings. In order to make lookup table stable, the given list of nodes are sorted and deduplicated.
func (*MaglevHash) Node ¶
func (m *MaglevHash) Node(key []byte) string
Node returns the assigned node for the given key.