Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Professional Cloud Database Engineer practice bank has an additional 769 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
A retail company runs an online order-processing application backed by a single Cloud SQL for MySQL primary instance provisioned for peak daytime traffic. The workload has predictable diurnal patterns—heavy read and write traffic during business hours and minimal activity overnight—along with steady annual growth and scheduled holiday flash sales.
The engineering team is encountering several challenges:
Which database scaling and pricing strategy should the team adopt to eliminate vertical scaling bottlenecks while optimizing cost efficiency?
Spanner is a globally distributed, strongly consistent relational database service that scales horizontally across compute nodes or processing units. Its managed autoscaler dynamically adds or removes compute capacity based on high-priority CPU utilization, total CPU utilization, and storage utilization targets within configured minimum and maximum limits.
1000 processing units, or 1 node, for autoscaling) and asymmetric read-only autoscaling to tune replica regions independently.Compared to statically provisioned vertical scaling, manual sharding, or average-load provisioning, Spanner's managed autoscaler uniquely combines horizontal read and write scalability with elastic, demand-driven pricing. It eliminates the operational overhead of sharding while preventing both overnight over-provisioning waste and daytime performance degradation.
Database sharding partitions data across multiple independent primary database instances—using a partitioning key such as customer ID or region—so that both reads and writes scale horizontally. Applying Committed Use Discounts (CUDs) lowers the hourly compute rate across these provisioned instances over a 1-year or 3-year term.
The organization were required to remain strictly on MySQL, experienced steady 24/7 traffic across evenly distributed tenants, and had dedicated engineering resources to build and maintain custom sharding logic.
Manual sharding achieves horizontal write scaling at the expense of application complexity and static over-provisioning. Managed distributed databases such as Spanner handle partitioning, data rebalancing, distributed transactions, and elastic autoscaling natively at the service layer.
Vertical scaling increases the compute capacity (vCPU and RAM) of a single database VM by upgrading to a larger machine type. Committed Use Discounts (CUDs) provide reduced hourly billing rates in exchange for a 1-year or 3-year commitment to continuous resource usage, while automatic storage increases dynamically expand disk space when available storage runs low.
The workload had steady, 24/7 predictable compute utilization comfortably within single-VM limits, and the primary goal was lowering the hourly rate of constant infrastructure while preventing unexpected disk exhaustion.
While CUDs are an effective billing optimization for stable baseline capacity, they cannot solve architectural vertical scaling ceilings or provide the cost savings of elastic scale-down during idle hours.
Downsizing a primary instance reduces its provisioned vCPU and RAM to match average daily traffic instead of peak demand, lowering fixed compute costs. Read replicas are asynchronous copies of the primary instance that serve read-only queries (SELECT statements) on separate compute instances.
INSERT, UPDATE, and DELETE operations must still execute on the single primary instance, leaving daytime write bottlenecks unresolved.The workload were overwhelmingly read-heavy with minimal write activity, allowing the downsized primary instance to comfortably absorb peak writes while read replicas handled daytime read surges.
A common misconception is that adding read replicas provides general horizontal scaling for relational databases. In single-primary architectures, read replicas address read capacity only; horizontal write scaling requires either a distributed database like Spanner or application-level sharding.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated