Hello,
The ScyllaDB team is pleased to announce the ScyllaDB Cloud Availability Zone Configuration.
Overview
ScyllaDB Cloud now offers you explicit control over Availability Zone placement when creating clusters. You can choose which Availability Zone IDs your nodes are deployed into, and optionally deploy to a single Availability Zone for reduced cost.
What’s New
Availability Zone ID Selection at Cluster Creation
When creating a new cluster, you can now specify the exact Availability Zone(s) for your nodes using Availability Zone IDs (e.g., use1-az1, use1-az2, use1-az3).
Previously, ScyllaDB Cloud automatically placed nodes into three availability zones. Now, Availability Zone placement is fully configurable via the ScyllaDB Cloud UI, Public API, and ScyllaDB Cloud Terraform Provider.
The user interface offers you the ability to select the availability zones. If the user chooses not to change the zones, they will be chosen at random as before.
With this change, we also updated ScyllaDB Cloud UI to always display and use the Availability Zone ID instead of names.
AWS Availability Zone Names (e.g., us-east-1a) are not consistent across accounts — the same name can map to a different physical zone depending on the user account.
On the other hand, the Availability Zone IDs are stable identifiers that refer to the same physical zone regardless of which AWS account is used.
This change affects only AWS, since GCP does not suffer from such an issue.
Multi-Availability Zone (Default)
Multi-Availability Zone deployment remains the default and recommended configuration for production clusters. 3 Availability Zones are recommended with one replica per zone.
Single Availability Zone Deployment
We introduced a new deployment option that allows you to place all cluster nodes in a single Availability Zone.
This reduces cost but concentrates all nodes in a single Availability Zone. In the rare event that the zone becomes unavailable, the cluster will be inaccessible.
Note: A single Availability Zone does not mean zero redundancy at the server level. ScyllaDB Cloud uses spread placement groups within the zone to distribute nodes across distinct physical hosts, helping protect against individual hardware failures. If the entire zone becomes unavailable, so does the cluster. Not recommended for production workloads.
A brief comparison between the two options.
| Single Availability Zone | Multi-Availability Zone (default) | |
|---|---|---|
| High Availability | ✗ No (zone failure = cluster unavailable) | ✓ Yes (survives zone failure) |
| Cost | Lower (no cross-zone data transfer) | Standard |
| Node failure protection | ✓ Spread placement groups | ✓ Distributed across zones |
| Recommended for | Dev, test, cost-sensitive workloads | Production |
More about ScyllaDB Cloud Availability Zone Configuration can be found in the Documentation.