From 7f166c33727742e87cf4660d7905b3f943706c5c Mon Sep 17 00:00:00 2001 From: Anon Ray Date: Wed, 8 May 2024 17:52:13 +0530 Subject: [PATCH] docs: add a para about random selection of read replicas PR-URL: https://github.com/hasura/graphql-engine-mono/pull/10807 GitOrigin-RevId: 986ccc2114e5dc4b67048091cc8358058b4a60e8 --- docs/docs/databases/database-config/read-replicas.mdx | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/docs/docs/databases/database-config/read-replicas.mdx b/docs/docs/databases/database-config/read-replicas.mdx index 981160d9209..9dda679947f 100644 --- a/docs/docs/databases/database-config/read-replicas.mdx +++ b/docs/docs/databases/database-config/read-replicas.mdx @@ -29,6 +29,11 @@ import ProductBadge from '@site/src/components/ProductBadge'; Hasura Cloud and Hasura Enterprise can route queries and subscriptions across _read replicas_ while sending all mutations and Metadata API calls to the primary. +This works by selecting a read replica instance at random. Hasura does not +perform any kind of load balancing across the replicas. The random function used +has a uniform distribution, and so over time, queries will distribute uniformly +across all replica instances. + ## Adding read replica URLs ### Postgres