What is Cloud Spanner Data Boost — how does it help?

Cloud Spanner may cause huge billings if it is not built as needed. Sometimes, you may have to create a resource-intensive spanner instance just to support some adhoc or periodical queries. Well, this will not be the case hereafter if you could leverage newly announced Data Boost features.

Description

• Cloud Spanner Data Boost is a fully managed, serverless service.
• It offers independent compute resources for supported Cloud Spanner workloads.

Workload Isolation:

• Data Boost allows executing analytics queries and data exports with minimal impact on existing workloads in the provisioned Spanner instance.
• It achieves this by utilizing separate Spanner clusters managed by Google at the region level.

Query Routing:

• For eligible queries requesting Data Boost, Spanner transparently routes the workload to dedicated servers.
• Eligible queries are those with the first operator in the query execution plan being a distributed union.
• Queries Do NOT require modifications to leverage Data Boost.

Impactful Scenarios:

• Data Boost is particularly effective in scenarios where resource contention should be avoided in the existing transactional system
• Ad hoc or infrequent queries processing large data volumes, like federated queries from BigQuery to Spanner.
• Reporting or data export tasks, such as Dataflow jobs exporting Spanner data to Cloud Storage.

Below is a diagram that illustrates how Data Boost coordinates with the Spanner instance to provide independent compute resources.

Summary of benefits with Data Boost:

• Data Boost ensures that running supported queries against the latest data won’t disrupt ongoing transactional work, regardless of the query’s complexity or the volume of data processed.
• It maintains, and in some cases even improves, the speed at which queries are processed, ensuring efficient performance.
• Data Boost prevents the need for over-provisioning of Spanner instances solely for occasional analytics queries, helping save on resources and costs.
• It offers exceptional scalability, allowing for increased query parallelism that can dynamically adapt to fluctuating workloads, making it highly responsive to sudden bursts of demand.
• Administrators gain access to comprehensive metrics, enabling them to pinpoint the most resource-intensive queries and identify cost drivers. They can then fine-tune optimizations and track the impact on serverless processing unit consumption in subsequent query executions.
• Implementing Data Boost requires no additional operational burdens. There’s no need to manage an extra service, engage in capacity planning, or handle provisioning. Scaling happens seamlessly, and maintenance is not a concern.

Currently offered in below locations:

• asia-northeast1 (Tokyo)
• us-central1 (Iowa)
• southamerica-east1 (São Paulo)
• europe-west1 (Belgium)
• europe-west2 (London)
• europe-west3 (Frankfurt)