ScyllaDB Cloud Best Practices¶

While Creating your cluster, make sure you enable VPC peering.

VPC peering is more secure, it will save network costs and it has no downsides.

Enabling this will also allow you future flexibility by using features that require this setting. Read more here.

When setting up ScyllaDB Cloud to run in your account (BYOA), make sure you set all your account limits with the needed additional allocations for ScyllaDB Cloud. See BYOA for AWS and BYOA for GCP for the needed resources and limits. When setting up ScyllaDB Cloud to run in your account (BYOA), you will not be able to use the BYOA option without confirming you have created the necessary allocations.

Over the lifetime of your cluster there may be cases where your server’s IP address changes or you may need to replace a node or two. As long as your drivers are live it should not be an issue because the driver will learn the new topology and adjust accordingly.

When setting up ScyllaDB Cloud to run in your account (BYOA), ScyllaDB Cloud creates resources within your account and assumes exclusive management of those resources. If you modify any infrastructure components managed by ScyllaDB Cloud, we cannot guarantee that the service will continue to operate as expected. This includes modifications to security groups, permissions associated with the ScyllaDB role, or other restrictions that could disrupt ScyllaDB Cloud’s ability to manage the necessary resources. Such changes may affect our ability to uphold service commitments within your account.

To avoid service disruptions, please consult our support team before making any changes to ScyllaDB-managed resources.

Always choose NetworkTopology as Replication Strategy for all your Keyspaces. Using SimpleStrategy will not provide High Availability in case of an Availability Zone (AZ) failure, and will not allow multiple Data Centers(DC) setup.

Read more in ScyllaDB Docs or Scylla University

Use Consistency Level of Quorum for one DC, or Local Quorum for Multi DC cluster.

When creating your Keyspaces, make sure to use the same Replication Factor (RF) you used for creating the cluster. A lower number will reduce the cluster availability, while a high number may place more than once replica on the same zone.

When sizing your cluster, make sure you have at least 50% free disk space for cluster backup and other maintenance routines (e.g., compaction).