1. General-purpose
   1. These have the best price-performance with the most flexible vCPU to memory ratios
   2. Series
      1. E2
         1. Workload
            1. (Cost-Optimized) Day-to-day computing at a lower cost
            2. Intel processor or the 2nd Gen AMD EPYC Rome processor with up to 32 vCPUs with up to 128 GB of memory with a maximum of 8 GB per vCPU
            3. E2 contain shared-core machine types with 2 vCPUs for short periods of bursting
         2. Applications
            1. Low-traffic web servers
            2. Back office apps
            3. Containerized microservices
            4. Microservices
            5. Virtual desktop
            6. Development and test environments
      2. N2, N2D, N1
         1. Workload
            1. (Balanced) Balanced price/performance across a wide range of VM shapes
            2. N2 are Intel-based VMs (Ice Lake and Cascade Lake) with up to 128 vCPUs and 0.5 to 8 GB of memory per vCPU
            3. N2D are AMD-based VMs (EPYC Rome 2nd Gen and EPYC Milan 3rd Gen) with up to 224 vCPUs and 8 GB of memory per vCPU
            4. N1 are Intel-based VMs (Sandy Bridge, Ivy Bridge, Broadwell, and Skylake) with up to 96 vCPUs and 6.5 GB of memory per vCPU N1 offers f1-micro and g1-small shared-core machines types which have up to 1 vCPU available for short periods of bursting
         2. Applications
            1. Low to medium traffic web and app servers
            2. Containerized microservices
            3. Business intelligence apps
            4. Virtual desktops
            5. CRM applications
            6. Data Pipelines
      3. C3, C3D
         1. Workload
            1. Consistently high performance for a variety of workloads
            2. C3 are Intel-based VMs (Sapphire Rapids and Google's custom Intel Infrastructure Processing Unit) with up to 176 vCPUs and 2, 4, or 8 GB of memory per vCPU
            3. C3D are AMD-based VMs (EPYC Genoa and Google's custom Intel Infrastructure Processing Unit) with up to 360 vCPUs and 2, 4, or 8 GB of memory per vCPU
         2. Applications
            1. High traffic web and app servers
            2. Databases
            3. In-memory caches
            4. Ad servers
            5. Game servers
            6. Data analytics
            7. Media streaming and transcoding
            8. CPU-based ML training and inference
      4. Tau T2D, Tau T2A
         1. Workload
            1. Best per-core performance/cost for scale-out workloads
            2. Each Tau T2D VM can have up to 60 vCPS, 4 GB of memory per vCPU, and is available on 3rd Gen AMD EPYC Milan processors.
         2. Applications
            1. Scale-out workloads
            2. Web serving
            3. Containerized microservices
            4. Media transcoding
            5. Large-scale Java applications
2. Compute-optimized
   1. These have the highest performance per core on Compute Engine and is optimized for compute-intensive workloads
   2. Series
      1. H3, C2, C2D
         1. Workload
            1. Ultra high performance for compute-intensive workloads
            2. H3 VMs offer 88 vCPUs and 352 GB of DDR5 memory. H3 VMs run on Intel Sapphire Rapids CPU platform and Google's custom Intel Infrastructure Processing Unit (IPU)
            3. C2 VMs offer up to 60 vCPUs, 4 GB of memory per vCPU, and are available on the Intel Cascade Lake CPU platform
            4. C2D VMs offer up to 112 vCPUs, up to 8 GB of memory per vCPU, and are available on the 3rd Gen AMD EPYC Milan platform
         2. Applications
            1. Compute-bound workloads
            2. High-performance web servers
            3. Game servers
            4. High performance computing (HPC)
            5. Media transcoding
            6. Modeling and simulation workloads
            7. AI/ML
3. Memory-optimized
   1. Provides the most compute and memory resources of any Compute Engine machine family offering.
   2. Series
      1. M3, M2, M1
         1. Workload
            1. Highest memory to compute ratios for memory-intensive workloads
            2. M1 VMs offer up to 160 vCPUs, 14.9 GB to 24 GB of memory per vCPU, and are available on the Intel Skylake and Broadwell CPU platform
            3. M2 VMs are available as 6 TB, 9 TB, and 12 TB machine types, and are available on Intel Cascade Lake CPU platform
            4. M3 VMs offer up to 128 vCPUs, with up to 30.5 GB of memory per vCPU, and are available on the Intel Ice Lake CPU platform
         2. Applications
            1. Medium to extra-large SAP HANA in-memory databases
            2. In-memory data stores, such as Redis
            3. Simulation
            4. High performance databases such as Microsoft SQL Server, MySQL
            5. Electronic design automation
4. Accelerator-optimized
   1. Is ideal for massively parallelized Compute Unified Device Architecture (CUDA) compute workload, such as ML and HPC
   2. Series
      1. A2, G2
         1. Workload
            1. Optimized for accelerated high performance computing workload
            2. A2 VMs offer 12 to 96 vCPUs, up to 1360 GB of memory, and are available on the Intel Cascade Lake CPU platform
            3. G2 VMs offer 4 to 96 vCPUs, up to 432 GB of memory, and are available on the Intel Cascade Lake CPU platform
         2. Applications
            1. CUDA-enable ML training and interference
            2. High-performance computing (HPC)
            3. Massively parallelized computing
            4. BERT natural language processing
            5. Deep learning recommendation model (DLRM)
            6. Video transcoding
            7. Remote visualization workstation