Skip to content

Add availability zone awareness to node based routing#6189

Merged
nadav-govari merged 4 commits intonadav/feature-node-based-routingfrom
nadav/az-aware
Mar 9, 2026
Merged

Add availability zone awareness to node based routing#6189
nadav-govari merged 4 commits intonadav/feature-node-based-routingfrom
nadav/az-aware

Conversation

@nadav-govari
Copy link
Collaborator

Description

Uses the already broadcasted cluster-level property availability_zone to inform ingest decisions. Similar to the old shard based routing table, splits candidate nodes into local/remote vecs, and prefers to pick from same-AZ nodes before other-AZ nodes.

This assumes that there is an equal distribution of traffic between AZs. If not, the AZ that sees more traffic will get hotter sooner, and will create a hotspot. For now, it's up to users to ensure that their traffic is load balanced across the cluster properly.

How was this PR tested?

Unit tests. end to end tests incoming.

Base automatically changed from nadav/pr4 to nadav/feature-node-based-routing March 3, 2026 19:10
@nadav-govari nadav-govari merged commit 1d0e18e into nadav/feature-node-based-routing Mar 9, 2026
4 checks passed
@nadav-govari nadav-govari deleted the nadav/az-aware branch March 9, 2026 18:47
nadav-govari added a commit that referenced this pull request Mar 16, 2026
* Implement IngesterCapacityScore broadcast (#6152)

* Implement node based routing table (#6159)

* Use new node based routing table for routing decisions (#6163)

* Piggyback routing update on persist response (#6173)

* Remove unused shard_ids in persist protos (#6169)

* Add availability zone awareness to node based routing (#6189)

* Remove old routing table; Take both disk and memory WAL readings (#6193)

* Add az-aware ingest attempts metric (#6194)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants