aboutsummaryrefslogtreecommitdiff
path: root/src/table/replication/sharded.rs
blob: 966be31a3950c81f4ed3622cc8e3622003a8df17 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
use std::sync::Arc;

use garage_rpc::membership::System;
use garage_rpc::ring::Ring;
use garage_util::data::*;

use crate::replication::*;

#[derive(Clone)]
pub struct TableShardedReplication {
	pub system: Arc<System>,
	pub replication_factor: usize,
	pub read_quorum: usize,
	pub write_quorum: usize,
}

impl TableReplication for TableShardedReplication {
	// Sharded replication schema:
	// - based on the ring of nodes, a certain set of neighbors
	//   store entries, given as a function of the position of the
	//   entry's hash in the ring
	// - reads are done on all of the nodes that replicate the data
	// - writes as well

	fn partition_of(&self, hash: &Hash) -> u16 {
		self.system.ring.borrow().partition_of(hash)
	}

	fn read_nodes(&self, hash: &Hash) -> Vec<UUID> {
		let ring = self.system.ring.borrow().clone();
		ring.walk_ring(&hash, self.replication_factor)
	}
	fn read_quorum(&self) -> usize {
		self.read_quorum
	}

	fn write_nodes(&self, hash: &Hash) -> Vec<UUID> {
		let ring = self.system.ring.borrow();
		ring.walk_ring(&hash, self.replication_factor)
	}
	fn write_quorum(&self) -> usize {
		self.write_quorum
	}
	fn max_write_errors(&self) -> usize {
		self.replication_factor - self.write_quorum
	}

	fn split_points(&self, ring: &Ring) -> Vec<Hash> {
		let mut ret = vec![];

		for entry in ring.ring.iter() {
			ret.push(entry.location);
		}
		if ret.len() > 0 {
			assert_eq!(ret[0], [0u8; 32].into());
		}

		ret
	}
}