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.
You are migrating a business-critical, read-heavy MySQL 8.4 (InnoDB) OLTP database to Compute Engine. Load testing reveals the following workload characteristics and requirements:
InnoDB buffer pool to minimize disk readsDuring initial testing on a 16-vCPU standard machine type (approximately 60 GB of RAM), frequent disk reads degraded read throughput.
Which machine type configuration should you select?
standard predefined machine type in the C4 or C3D series to reach 120–128 GB of RAM.standard machine types arestandard machine types offer a balanced compute-to-memory ratio of approximately 3.75 to 4 GB of RAM per vCPU (3.75 GB/vCPU on C4 and 4 GB/vCPU on C3D). They are designed as a general baseline for mixed workloads where CPU utilization and memory consumption scale proportionally.
standard instance provides the 120–128 GB of RAM needed for the 100 GB working set, it forces you to double your compute allocation from 16 to 32 vCPUs.highmem: Requires twice as many vCPUs as a highmem shape to achieve the same InnoDB buffer pool capacity.A 32-vCPU standard machine type would be the right choice if monitoring showed that both CPU utilization and memory were saturated at 16 vCPUs—for example, if heavy query execution, sorting, encryption, or connection handling required 32 vCPUs alongside 120–128 GB of RAM.
A common capacity-planning mistake is scaling a standard VM vertically solely to resolve memory pressure. When a database is memory-bound rather than CPU-bound, shifting horizontally across shape ratios (standard to highmem) provides the required RAM at a substantially lower total cost.
highcpu predefined machine type (approximately 2 GB of RAM per vCPU) with 60–64 vCPUs to provide 120–128 GB of RAM and higher concurrency.highcpu machine types arehighcpu machine types prioritize compute density over memory capacity, providing approximately 2 GB of RAM per vCPU (such as 2 GB/vCPU on C4 and roughly 2 GB/vCPU on C3D and C2D). They are engineered for compute-bound workloads that perform heavy mathematical, analytical, or parallel processing without needing large in-memory caches.
highcpu configurations carry only ~2 GB of RAM per vCPU, reaching 120–128 GB of memory requires provisioning 60 to 64 vCPUs—nearly four times the compute capacity of the 16-vCPU baseline.InnoDB buffer pool.highmem shapes.highcpu shape escalates compute costs four times faster per gigabyte of RAM than scaling on a highmem shape.A highcpu configuration would be appropriate if the database had a very small working set (for example, under 16–32 GB) but suffered from intense CPU-bound bottlenecks such as complex analytical computations, stored procedures, or massive concurrent query parsing.
Database engineers sometimes select highcpu shapes assuming higher concurrency automatically improves OLTP throughput. However, when the InnoDB buffer pool cannot hold the working set, extra vCPUs simply sit idle waiting on disk I/O rather than accelerating queries.
N4 machine type with minimal vCPUs and extended memory to provide 128 GB of RAM at the lowest vCPU count.N4 machine types with extended memory areCustom machine types in the N4 general-purpose series allow you to independently specify vCPU and memory counts, and extended memory allows exceeding the default memory-to-vCPU ceiling (8 GB per vCPU). N4 instances use dynamic resource management (DRM) to optimize infrastructure cost for general-purpose workloads.
InnoDB lock contention, and handle client connections.C4/C4A performance guarantees: N4 relies on dynamic resource management rather than the dedicated, consistent, low-latency guarantees provided by C4 and C4A for business-critical databases.highmem instances.N and E series VMs, not on C4 or C4A.An N4 custom machine type with extended memory would be optimal if the database had a large working set (100+ GB) paired with very low query rates (for example, infrequent reporting or low-traffic internal tools) where strict latency guarantees and high Hyperdisk IOPS were unnecessary.
While minimizing vCPUs via extended memory reduces raw CPU spend, high-throughput OLTP systems require a balanced ratio (~8 GB RAM per vCPU on highmem) to sustain query execution, network traffic, and block storage I/O ceilings.
highmem predefined machine type in the C4 or C4A series (approximately 8 GB of RAM per vCPU, providing 124–128 GB of total RAM).highmem predefined machine types arePredefined highmem machine types in Compute Engine provide roughly 8 GB of RAM per vCPU (7.75 GB per vCPU on C4 and 8 GB per vCPU on C4A). The C4 (Intel Xeon) and C4A (Arm-based Google Axion) series belong to the latest generation of performance-focused general-purpose machines powered by Google Titanium, delivering predictable, low-latency compute and storage performance.
InnoDB buffer pool to hold the entire 100 GB working set while leaving 24–28 GB for the operating system, active connections, and internal database buffers.C4 or C4A provides strong per-core performance, hardware isolation, and sufficient core count to support high Google Cloud Hyperdisk throughput and IOPS limits.standard) to ~8 GB/vCPU (highmem) doubles available RAM without increasing the vCPU count or incurring custom machine type premiums.InnoDB buffer pool dramatically increases read QPS compared to disk-bound queries.C4 and C4A instances provide dedicated vCPU resources, advance host maintenance notifications, and high network/storage ceilings.highmem shapes.Compared to scaling up a standard or highcpu shape, choosing the smallest highmem instance that fits the working set avoids paying for idle vCPUs. Compared to a low-vCPU custom N4 instance with extended memory, a 16-vCPU C4 or C4A highmem machine preserves the compute concurrency, consistent latency guarantees, and Hyperdisk performance limits needed for a business-critical, high-QPS OLTP workload.
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