Cluster Types - X Cloud and Standard¶

X Cloud Auto Scaling¶

X Cloud clusters scale automatically using different instance families and a mix of instance sizes, increasing or decreasing new instances as the workload spikes or drops. This brings multiple benefits to the scaling process.

• You avoid overprovisioning. The cluster capacity is automatically adjusted to match real-time demand.

• It enables up to 90% storage utilization, allowing you to maintain reliable performance at the lowest possible cost.

• Scaling is faster. Due to the tablets-based replication architecture, new nodes can be added in parallel and almost instantly balance the load.

X Cloud clusters are auto-scaled within one predefined instance family, but with different instance types to ensure optimal storage utilization. See Scaling Policy for details.

Auto Scaling Policy¶

When you create an X Cloud cluster, you can define an initial auto-scaling policy. You can update that policy later as needed.

To define the initial auto-scaling policy when creating a new cluster, configure the options in Data Center Scaling Policy.

• Instance Family - Specifies the instance family to be used for auto-scaling. X Cloud clusters are auto-scaled within one predefined instance family. The default is i3en.

• Instance Type - Specifies instance types to be used for auto-scaling. By default, all instance types within a given family can be used when auto-scaling a cluster to ensure maximum storage utilization.

• Min vCPU Count - Specifies the minimum number of vCPUs to reserve compute capacity for that particular cluster. You can customize it depending on the initial load you anticipate for your cluster. If set, ScyllaDB Cloud will never scale the cluster below this vCPU count, even during automatic downscaling operations. The default is 0.

• Min Storage Size - Specifies the minimum physical storage capacity. You can customize it depending on the initial load you anticipate for your cluster. If set, ScyllaDB Cloud will never scale the cluster below this capacity, even during automatic downscaling operations. The default is 0.

• Target Storage Utilization - Specifies the storage utilization you want to maintain. ScyllaDB will be performing auto-scaling operations with the goal of maintaining the storage utilization level specified by this option.

  • The default is 80%.

  • The maximum is 90%.

  • You should adjust the value depending on the workload. For workloads with a high volume of write operations, we recommend values lower than 85 to ensure that the ScyllaDB Cloud will be able to scale the cluster.

  • In clusters with a small total storage capacity, storage utilization may vary from the configured target due to instance size granularity.

To modify the auto-scaling policy for an existing cluster, go to your cluster’s page and edit the X CLOUD MINIMAL SCALING POLICY.

After you customize the auto-scaling policy, the deployment information will be displayed.

The Data Center summary will show:

• Physical Storage - The total initial amount of physical storage allocated to all nodes in the cluster.

• Storage available for data - The space available for storing your data. The calculation is based on two key factors:

  • Replication Factor (RF), which determines the number of data replicas stored. For example, with an RF of 3, your data occupies storage space on three separate nodes.

  • Estimated storage-level compression. Storage-level compression is applied to data within SSTables to reduce the space used on the physical storage.

• vCPU - The number of vCPUs.

• Availability Zones - The number of availability zones.

The Instances Per Zone summary will show the instance type configuration for each zone. The cluster consists of 3 identical zones, automatically replicated.

X Cloud Limitations¶

X Cloud clusters use tablet-based data distribution. Some ScyllaDB features are not yet supported for tablets, and therefore are currently unavailable in X Cloud clusters. See Tablets Limitations.

We are actively working to add support for these features in future updates.