aboutsummaryrefslogtreecommitdiff
path: root/doc/Load_Balancing.md
diff options
context:
space:
mode:
authorAlex Auvolat <alex@adnab.me>2021-02-25 11:27:13 +0100
committerAlex Auvolat <alex@adnab.me>2021-02-25 11:27:13 +0100
commit0522983aec2d7daccb47160bfcdf6c09c048a35e (patch)
treeeb470acd7e288086aca34d0a41fa4f78f950160e /doc/Load_Balancing.md
parentfdf908e8452fc57291c03eb85aa63de5c1b8093d (diff)
downloadgarage-0522983aec2d7daccb47160bfcdf6c09c048a35e.tar.gz
garage-0522983aec2d7daccb47160bfcdf6c09c048a35e.zip
precisions
Diffstat (limited to 'doc/Load_Balancing.md')
-rw-r--r--doc/Load_Balancing.md12
1 files changed, 7 insertions, 5 deletions
diff --git a/doc/Load_Balancing.md b/doc/Load_Balancing.md
index 808bb4b3..1d508fa0 100644
--- a/doc/Load_Balancing.md
+++ b/doc/Load_Balancing.md
@@ -8,7 +8,7 @@ I have conducted a quick study of different methods to load-balance data over di
- *minimal disruption*: when adding or removing a node, as few partitions as possible should have to move around
-- *order-agnostic*: the same set of nodes (associated with a datacenter name
+- *order-agnostic*: the same set of nodes (each associated with a datacenter name
and a capacity) should always return the same distribution of partition
replicas, independently of the order in which nodes were added/removed (this
is to keep the implementation simple)
@@ -19,10 +19,12 @@ I have conducted a quick study of different methods to load-balance data over di
This strategy can be used with any ring-like algorithm to make it aware of the *multi-datacenter* requirement:
-- the ring is a list of positions, each associated with a single node in the cluster
-- look up position of item on ring
-- select the node for that position
-- go clockwise, skipping nodes that:
+In this method, the ring is a list of positions, each associated with a single node in the cluster.
+Partitions contain all the keys between two consecutive items of the ring.
+To find the nodes that store replicas of a given partition:
+
+- select the node for the position of the partition's lower bound
+- go clockwise on the ring, skipping nodes that:
- we halve already selected
- are in a datacenter of a node we have selected, except if we already have nodes from all possible datacenters