When you create a Compute Engine VM, one of the first decisions you make is the machine type. The machine type determines how many virtual CPUs and how much memory the VM has, and it also determines the underlying hardware and pricing. Google Cloud organizes machine types into families, each optimized for a different category of workload. For the Associate Cloud Engineer exam, you need to understand the naming convention, know the major families, and be able to identify which family fits a described scenario.
Compute Engine machine type names follow a consistent pattern: family-workload-vcpus. For example, n2-standard-4 means N2 family, standard workload profile, 4 vCPUs. The e2-micro name means E2 family, micro tier, with shared vCPUs rather than dedicated ones.
The number at the end of the name is almost always the vCPU count. n1-standard-32 has 32 vCPUs. n2-highcpu-16 has 16 vCPUs and is configured with a higher ratio of CPU to memory than the standard tier. n2-highmem-8 has 8 vCPUs and a higher ratio of memory to CPU than standard.
The memory allocation is determined by the family and the workload tier. Standard machines have a balanced CPU-to-memory ratio. High-memory machines have more RAM per vCPU. High-CPU machines have less RAM per vCPU. This lets you tune the machine type to your specific workload without over-provisioning in one dimension.
E2 machines are the most cost-efficient general-purpose machines on Compute Engine. They run on shared physical cores, which means you do not have dedicated access to the underlying hardware - your vCPUs share time with other VMs. This makes them cheaper, but it also means performance is more variable under sustained load.
E2 machines are appropriate for development environments, low-traffic web servers, small databases, and internal tools where performance predictability is not critical. The e2-micro and e2-small machine types are among the cheapest VMs available and are commonly used for lightweight services or as part of the Google Cloud Free Tier.
N2 machines are the current generation of balanced general-purpose machines. They offer dedicated vCPUs, consistent performance, and good cost efficiency for most production workloads. N2 machines are the right default choice when you need reliable performance for web applications, API servers, or data processing pipelines that need consistent throughput.
N1 machines are the previous generation. They are still available and used in many existing deployments, but N2 offers better performance per dollar for new deployments. When the exam mentions n1-standard-1 or n1-standard-4, it is referring to the older N1 generation.
C2 machines are designed for compute-intensive workloads that need maximum CPU performance. They offer the highest per-core performance of any Compute Engine family and are appropriate for scientific modeling, game servers, high-performance computing, and anything where raw CPU throughput is the bottleneck.
C2 machines are more expensive than N2 for the same number of vCPUs, so they should only be chosen when the workload genuinely needs the additional CPU performance. Using a C2 for a low-traffic web server would be significant over-provisioning.
M1 and M2 machines are built for workloads that need very large amounts of RAM. These are used for in-memory databases like SAP HANA, large-scale data analytics that benefit from keeping data in memory, and any application that needs more memory than the high-memory tiers of the standard families can provide.
Memory-optimized machines have very high RAM-to-vCPU ratios. An m2-ultramem-416 has 416 vCPUs and nearly 12 TB of memory. These are specialized machines for specific enterprise workloads, and the exam does not expect you to know the specific specs - just that M-series machines are for memory-intensive workloads.
Beyond the predefined machine types, Compute Engine lets you create custom machine types where you specify exactly how many vCPUs and exactly how much memory you want. This is useful when the predefined options do not match your workload well - for example, if you need more memory than the standard tier provides but less than the high-memory tier, a custom type lets you specify the exact amount.
Custom machine types are available in the N1, N2, and E2 families. They are billed based on the vCPU and memory resources you select, and the pricing is slightly higher per vCPU and per GB of memory than the equivalent predefined type.
The Associate Cloud Engineer exam presents machine type questions as scenario-based matching problems. A description of a workload maps to a machine type family. Some common patterns:
A cost-sensitive development environment with variable traffic - E2, because shared vCPUs are acceptable and the cost savings matter. A production web application that needs consistent performance - N2, because it is the current general-purpose generation with dedicated vCPUs. A scientific simulation that needs maximum CPU throughput - C2, compute-optimized. A large in-memory analytics database - M1 or M2, memory-optimized.
The exam also tests the naming convention directly. If a question asks how many vCPUs n2-standard-8 has, the answer is eight - the last number in the name is the vCPU count.
My Associate Cloud Engineer course covers machine types in the Compute Engine section alongside instance configuration, pricing, and the other VM settings you need to understand for the Associate Cloud Engineer exam.
