Create a VM with a custom machine type

Stay organized with collections Save and categorize content based on your preferences.

Compute Engine offers predefined machine types that you can use when you create a VM instance. A predefined machine type has a preset number of vCPUs and amount of memory, and is charged at a set price.

If predefined VMs don't meet your needs, you can create a VM instance with custom virtualized hardware settings. Specifically, you can create a VM instance with a custom number of vCPUs and amount of memory, effectively using a custom machine type. Custom machine types are available in the general-purpose machine family. When you create a custom VM, you're deploying a custom VM from the E2, N2, N2D, or N1 machine family.

Custom VMs are ideal in the following scenarios:

  • Workloads that aren't a good fit for the predefined virtual machine types.
  • Workloads that require more processing power or more memory but don't need all upgrades that are provided by the next level machine type.

Before you begin

Pricing

Google charges for custom VMs based on the number of vCPUs and memory hours that the VM uses. This is different from how predefined machine types are charged. See VM pricing for details.

Custom VMs are subject to the same 1-minute minimum charge as any other instance, but sustained use discounts for custom machine types are calculated differently. For more information, see sustained use discounts for custom VMs.

Expressing memory in GB or MB

For Google Cloud tools and documentation, machine type memory is calculated in gigabytes (GB), where 1 GB is 230 bytes. This unit of measurement is also known as a gibibyte (GiB) . When converting memory from GB to MB, 1 GB = 1024 MB.

In the API, you must always provide memory in megabytes. If you use the Google Cloud CLI, you can provide the total memory for a VM in gigabytes or megabytes. However, the gcloud CLI expects the memory value to be an integer, so you cannot provide a float value. For example, to express 5.75 GB, convert 5.75 GB into MB instead. In this case, 5.75 GB is 5888 MB.

Create a VM with a custom machine type

Before you create a VM instance, make sure you read and understand the custom specifications for creating this machine type.

Console

  1. In the Google Cloud console, go to the Create an instance page.

    Go to Create an instance

  2. In the Zone list, select the zone where you want to host this VM. The Series list is filtered to include only the machine type families available in the selected zone.

  3. In the Machine configuration section, select General-purpose.

    1. In the Series list, click N1 in First Generation for N1 custom machine types or E2, N2, or N2D for Second Generation custom machine types.
    2. From the Machine type list, select Custom.
    3. To specify the number of vCPUs and the amount of memory for the VM instance, drag the sliders or enter the values in the text boxes. The console displays an estimated cost for the VM as you change the number of vCPUs and memory.
  4. Continue to create the VM.

gcloud

The gcloud command for custom machine types differs slightly depending on the machine type family.

For N1 machine types, use the gcloud compute instances create command and include one of the following options:

  • The --custom-cpu and --custom-memory flags.
  • The --machine-type=custom-[NUMBER_OF_CPUS]-[NUMBER_OF_MB] flag.

For example, the following command would create an instance running an N1 machine type with 4 vCPUs and 5 GB of total memory:

gcloud compute instances create example-instance \
    --custom-cpu=4 --custom-memory=5

For N2 machine types, use the gcloud compute instances create command and include one of the following options:

  • The --custom-cpu, --custom-memory, and --custom-vm-type flags.
  • The --machine-type=n2-custom-NUMBER_OF_CPUS-NUMBER_OF_MB flag.

When using the --custom-memory flag, specify the total amount of memory in GB or in MB. The property must be an integer, so if you want to specify increments of 0.25 GB for memory, convert that value to MB instead.

To specify an incremental value, such as 2.5 GB of memory, convert the value to MB and supply the value followed by the MB suffix. The following example creates an instance running an N2 machine type:

gcloud compute instances create example-instance \
    --custom-cpu=6 --custom-memory=3072MB --custom-vm-type=n2

Alternatively, you can specify a custom machine type using the following format:

gcloud compute instances create example-instance \
    --machine-type n2-custom-NUMBER_OF_CPUS-NUMBER_OF_MB

Replace the following:

  • NUMBER_OF_CPUS: The number of CPUs you want.
  • NUMBER_OF_MB: The amount of memory, in MB.

For example:

gcloud compute instances create example-instance --machine-type=n2-custom-6-3072

For N2D or E2 machine types, use the gcloud compute instances create command and include one of the following options:

  • The --custom-cpu, --custom-memory, and --custom-vm-type flags.
  • The --machine-type=n2d-custom-NUMBER_OF_CPUS-NUMBER_OF_MB flag.
  • The --machine-type=e2-custom-NUMBER_OF_CPUS-NUMBER_OF_MB flag.

For E2 shared-core custom machine types, use the same gcloud compute instances create command and include the shared-core machine size: micro, small, or medium.

  • The --machine-type=e2-custom SHARED_CORE_MACHINE_SIZE-NUMBER_OF_MB flag.

For example:

gcloud compute instances create example-instance \
    --machine-type=e2-custom-medium-NUMBER_OF_MB

When using the --custom-memory flag, specify the total amount of memory in GB or in MB. The property must be an integer. So, if you want to specify increments of 0.25 GB for memory, convert the value to MB instead.

To specify an incremental value, such as 2.5 GB of memory, convert the value to MB and supply the value followed by the MB suffix. The following example creates an instance running an E2 machine type:

gcloud compute instances create example-instance \
    --custom-cpu=6 --custom-memory=3072MB --custom-vm-type=e2

Alternatively, you can specify a custom machine type by using the following format:

gcloud compute instances create example-instance \
    --machine-type=e2-custom-NUMBER_OF_CPUS-NUMBER_OF_MB

Replace the following:

  • NUMBER_OF_CPUS: The number of CPUs you want.
  • NUMBER_OF_MB: The amount of memory, in MB.

For example:

gcloud compute instances create example-instance --machine-type=e2-custom-6-3072

API

In the API, create your instance creation request as you would normally, but when specifying a machineType value, use one of the following formats. Replace NUMBER_OF_CPUS with the number of CPUs and AMOUNT_OF_MEMORY_MB with the amount of memory you want. Provide memory in MB units.

  • For N1 machine types, use:

    zones/ZONE/machineTypes/custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY_MB
    
  • For N2 machine types, use:

    zones/ZONE/machineTypes/n2-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY_MB
    
  • For N2D machine types, use:

    zones/ZONE/machineTypes/n2d-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY_MB
    
  • For E2 machine types, use:

    zones/ZONE/machineTypes/e2-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY_MB
    
  • For E2 shared-core custom machine types, use:

    zones/ZONE/machineTypes/e2-custom-SHARED_CORE_MACHINE_SIZE-AMOUNT_OF_MEMORY_MB
    

    For example, the following URL would create an instance with 4 vCPUs and 5 GB (5120 MB) of memory:

    zones/us-central1-f/machineTypes/custom-4-5120

Go

import (
	"context"
	"fmt"
	"io"

	compute "cloud.google.com/go/compute/apiv1"
	computepb "google.golang.org/genproto/googleapis/cloud/compute/v1"
	"google.golang.org/protobuf/proto"
)

func customMachineTypeURI(zone, cpuSeries string, coreCount, memory int) (string, error) {
	const (
		n1       = "custom"
		n2       = "n2-custom"
		n2d      = "n2d-custom"
		e2       = "e2-custom"
		e2Micro  = "e2-custom-micro"
		e2Small  = "e2-custom-small"
		e2Medium = "e2-custom-medium"
	)

	type typeLimit struct {
		allowedCores     []int
		minMemPerCore    int
		maxMemPerCore    int
		allowExtraMemory bool
		extraMemoryLimit int
	}

	makeRange := func(start, end, step int) []int {
		if step <= 0 || end < start {
			return []int{}
		}
		s := make([]int, 0, 1+(end-start)/step)
		for start <= end {
			s = append(s, start)
			start += step
		}
		return s
	}

	containsString := func(s []string, str string) bool {
		for _, v := range s {
			if v == str {
				return true
			}
		}

		return false
	}

	containsInt := func(nums []int, n int) bool {
		for _, v := range nums {
			if v == n {
				return true
			}
		}

		return false
	}

	var (
		cpuSeriesE2Limit = typeLimit{
			allowedCores:  makeRange(2, 33, 2),
			minMemPerCore: 512,
			maxMemPerCore: 8192,
		}
		cpuSeriesE2MicroLimit  = typeLimit{minMemPerCore: 1024, maxMemPerCore: 2048}
		cpuSeriesE2SmallLimit  = typeLimit{minMemPerCore: 2048, maxMemPerCore: 4096}
		cpuSeriesE2MeidumLimit = typeLimit{minMemPerCore: 4096, maxMemPerCore: 8192}
		cpuSeriesN2Limit       = typeLimit{
			allowedCores:  append(makeRange(2, 33, 2), makeRange(36, 129, 4)...),
			minMemPerCore: 512, maxMemPerCore: 8192,
			allowExtraMemory: true,
			extraMemoryLimit: 624 << 10,
		}
		cpuSeriesN2DLimit = typeLimit{
			allowedCores:  []int{2, 4, 8, 16, 32, 48, 64, 80, 96},
			minMemPerCore: 512, maxMemPerCore: 8192,
			allowExtraMemory: true,
			extraMemoryLimit: 768 << 10,
		}
		cpuSeriesN1Limit = typeLimit{
			allowedCores:     append([]int{1}, makeRange(2, 97, 2)...),
			minMemPerCore:    922,
			maxMemPerCore:    6656,
			allowExtraMemory: true,
			extraMemoryLimit: 624 << 10,
		}
	)

	typeLimitsMap := map[string]typeLimit{
		n1:       cpuSeriesN1Limit,
		n2:       cpuSeriesN2Limit,
		n2d:      cpuSeriesN2DLimit,
		e2:       cpuSeriesE2Limit,
		e2Micro:  cpuSeriesE2MicroLimit,
		e2Small:  cpuSeriesE2SmallLimit,
		e2Medium: cpuSeriesE2MeidumLimit,
	}

	if !containsString([]string{e2, n1, n2, n2d}, cpuSeries) {
		return "", fmt.Errorf("incorrect cpu type: %v", cpuSeries)
	}

	tl := typeLimitsMap[cpuSeries]

	// Check whether the requested parameters are allowed.
	// Find more information about limitations of custom machine types at:
	// https://cloud.google.com/compute/docs/general-purpose-machines#custom_machine_types

	// Check the number of cores
	if len(tl.allowedCores) > 0 && !containsInt(tl.allowedCores, coreCount) {
		return "", fmt.Errorf(
			"invalid number of cores requested. Allowed number of cores for %v is: %v",
			cpuSeries,
			tl.allowedCores,
		)
	}

	// Memory must be a multiple of 256 MB
	if memory%256 != 0 {
		return "", fmt.Errorf("requested memory must be a multiple of 256 MB")
	}

	// Check if the requested memory isn't too little
	if memory < coreCount*tl.minMemPerCore {
		return "", fmt.Errorf(
			"requested memory is too low. Minimal memory for %v is %v MB per core",
			cpuSeries,
			tl.minMemPerCore,
		)
	}

	// Check if the requested memory isn't too much
	if memory > coreCount*tl.maxMemPerCore && !tl.allowExtraMemory {
		return "", fmt.Errorf(
			"requested memory is too large.. Maximum memory allowed for %v is %v MB per core",
			cpuSeries,
			tl.maxMemPerCore,
		)
	}
	if memory > tl.extraMemoryLimit && tl.allowExtraMemory {
		return "", fmt.Errorf(
			"requested memory is too large.. Maximum memory allowed for %v is %v MB",
			cpuSeries,
			tl.extraMemoryLimit,
		)
	}

	// Return the custom machine type in form of a string acceptable by Compute Engine API.
	if containsString([]string{e2Small, e2Micro, e2Medium}, cpuSeries) {
		return fmt.Sprintf("zones/%v/machineTypes/%v-%v", zone, cpuSeries, memory), nil
	}

	if memory > coreCount*tl.maxMemPerCore {
		return fmt.Sprintf(
			"zones/%v/machineTypes/%v-%v-%v-ext",
			zone,
			cpuSeries,
			coreCount,
			memory,
		), nil
	}

	return fmt.Sprintf("zones/%v/machineTypes/%v-%v-%v", zone, cpuSeries, coreCount, memory), nil
}

// createInstanceWithCustomMachineTypeWithHelper creates a new VM instance with a custom machine type.
func createInstanceWithCustomMachineTypeWithHelper(
	w io.Writer,
	projectID, zone, instanceName, cpuSeries string,
	coreCount, memory int,
) error {
	// projectID := "your_project_id"
	// zone := "europe-central2-b"
	// instanceName := "your_instance_name"
	// cpuSeries := "e2-custom-micro" // the type of CPU you want to use"
	// coreCount := 2 // number of CPU cores you want to use.
	// memory := 256 // the amount of memory for the VM instance, in megabytes.

	machineType, err := customMachineTypeURI(zone, cpuSeries, coreCount, memory)
	if err != nil {
		return fmt.Errorf("unable to create custom machine type string: %v", err)
	}

	ctx := context.Background()
	instancesClient, err := compute.NewInstancesRESTClient(ctx)
	if err != nil {
		return fmt.Errorf("NewInstancesRESTClient: %v", err)
	}
	defer instancesClient.Close()

	req := &computepb.InsertInstanceRequest{
		Project: projectID,
		Zone:    zone,
		InstanceResource: &computepb.Instance{
			Name: proto.String(instanceName),
			Disks: []*computepb.AttachedDisk{
				{
					InitializeParams: &computepb.AttachedDiskInitializeParams{
						DiskSizeGb: proto.Int64(10),
						SourceImage: proto.String(
							"projects/debian-cloud/global/images/family/debian-10",
						),
					},
					AutoDelete: proto.Bool(true),
					Boot:       proto.Bool(true),
				},
			},
			MachineType: proto.String(machineType),
			NetworkInterfaces: []*computepb.NetworkInterface{
				{
					Name: proto.String("global/networks/default"),
				},
			},
		},
	}

	op, err := instancesClient.Insert(ctx, req)
	if err != nil {
		return fmt.Errorf("unable to create instance: %v", err)
	}

	if err = op.Wait(ctx); err != nil {
		return fmt.Errorf("unable to wait for the operation: %v", err)
	}

	fmt.Fprintf(w, "Instance created\n")

	return nil
}

Java


import com.google.cloud.compute.v1.AttachedDisk;
import com.google.cloud.compute.v1.AttachedDiskInitializeParams;
import com.google.cloud.compute.v1.InsertInstanceRequest;
import com.google.cloud.compute.v1.Instance;
import com.google.cloud.compute.v1.InstancesClient;
import com.google.cloud.compute.v1.NetworkInterface;
import com.google.cloud.compute.v1.Operation;
import com.google.common.collect.ImmutableMap;
import java.io.IOException;
import java.util.Arrays;
import java.util.Collections;
import java.util.Map;
import java.util.Objects;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
import java.util.stream.IntStream;

public class CreateWithHelper {

  // This class defines the configurable parameters for a custom VM.
  static final class TypeLimits {

    int[] allowedCores;
    int minMemPerCore;
    int maxMemPerCore;
    int extraMemoryLimit;
    boolean allowExtraMemory;

    TypeLimits(int[] allowedCores, int minMemPerCore, int maxMemPerCore, boolean allowExtraMemory,
        int extraMemoryLimit) {
      this.allowedCores = allowedCores;
      this.minMemPerCore = minMemPerCore;
      this.maxMemPerCore = maxMemPerCore;
      this.allowExtraMemory = allowExtraMemory;
      this.extraMemoryLimit = extraMemoryLimit;
    }
  }

  public enum CpuSeries {
    N1("custom"),
    N2("n2-custom"),
    N2D("n2d-custom"),
    E2("e2-custom"),
    E2_MICRO("e2-custom-micro"),
    E2_SMALL("e2-custom-small"),
    E2_MEDIUM("e2-custom-medium");

    private static final Map<String, CpuSeries> ENUM_MAP;

    static {
      ENUM_MAP = init();
    }

    // Build an immutable map of String name to enum pairs.
    public static Map<String, CpuSeries> init() {
      Map<String, CpuSeries> map = new ConcurrentHashMap<>();
      for (CpuSeries instance : CpuSeries.values()) {
        map.put(instance.getCpuSeries(), instance);
      }
      return Collections.unmodifiableMap(map);
    }

    private final String cpuSeries;

    CpuSeries(String cpuSeries) {
      this.cpuSeries = cpuSeries;
    }

    public static CpuSeries get(String name) {
      return ENUM_MAP.get(name);
    }

    public String getCpuSeries() {
      return this.cpuSeries;
    }
  }

  // This enum correlates a machine type with its limits.
  // The limits for various CPU types are described in:
  // https://cloud.google.com/compute/docs/general-purpose-machines
  enum Limits {
    CPUSeries_E2(new TypeLimits(getNumsInRangeWithStep(2, 33, 2), 512, 8192, false, 0)),
    CPUSeries_E2MICRO(new TypeLimits(new int[]{}, 1024, 2048, false, 0)),
    CPUSeries_E2SMALL(new TypeLimits(new int[]{}, 2048, 4096, false, 0)),
    CPUSeries_E2MEDIUM(new TypeLimits(new int[]{}, 4096, 8192, false, 0)),
    CPUSeries_N2(
        new TypeLimits(concat(getNumsInRangeWithStep(2, 33, 2), getNumsInRangeWithStep(36, 129, 4)),
            512, 8192, true, gbToMb(624))),
    CPUSeries_N2D(
        new TypeLimits(new int[]{2, 4, 8, 16, 32, 48, 64, 80, 96}, 512, 8192, true, gbToMb(768))),
    CPUSeries_N1(
        new TypeLimits(concat(new int[]{1}, getNumsInRangeWithStep(2, 97, 2)), 922, 6656, true,
            gbToMb(624)));

    private final TypeLimits typeLimits;

    Limits(TypeLimits typeLimits) {
      this.typeLimits = typeLimits;
    }

    public TypeLimits getTypeLimits() {
      return typeLimits;
    }
  }

  static ImmutableMap<String, Limits> typeLimitsMap = ImmutableMap.<String, Limits>builder()
      .put("N1", Limits.CPUSeries_N1)
      .put("N2", Limits.CPUSeries_N2)
      .put("N2D", Limits.CPUSeries_N2D)
      .put("E2", Limits.CPUSeries_E2)
      .put("E2_MICRO", Limits.CPUSeries_E2MICRO)
      .put("E2_SMALL", Limits.CPUSeries_E2SMALL)
      .put("E2_MEDIUM", Limits.CPUSeries_E2SMALL)
      .build();

  // Returns the array of integers within the given range, incremented by the specified step.
  // start (inclusive): starting number of the range
  // stop (inclusive): ending number of the range
  // step : increment value
  static int[] getNumsInRangeWithStep(int start, int stop, int step) {
    return IntStream.range(start, stop).filter(x -> (x - start) % step == 0).toArray();
  }

  static int gbToMb(int value) {
    return value << 10;
  }

  static int[] concat(int[] a, int[] b) {
    int[] result = new int[a.length + b.length];
    System.arraycopy(a, 0, result, 0, a.length);
    System.arraycopy(b, 0, result, a.length, b.length);
    return result;
  }

  public static void main(String[] args)
      throws IOException, ExecutionException, InterruptedException, TimeoutException {
    // TODO(developer): Replace these variables before running the sample.
    // Project ID or project number of the Cloud project you want to use.
    String projectId = "your-google-cloud-project-id";
    // Name of the zone to create the instance in. For example: "us-west3-b".
    String zone = "google-cloud-zone";
    // Name of the new virtual machine (VM) instance.
    String instanceName = "instance-name";
    String cpuSeries = "N1";
    // Number of CPU cores you want to use.
    int coreCount = 2;
    // The amount of memory for the VM instance, in megabytes.
    int memory = 256;

    createInstanceWithCustomMachineTypeWithHelper(
        projectId, zone, instanceName, cpuSeries, coreCount, memory);
  }

  // Create a VM instance with a custom machine type.
  public static void createInstanceWithCustomMachineTypeWithHelper(
      String project, String zone, String instanceName, String cpuSeries, int coreCount, int memory)
      throws IOException, ExecutionException, InterruptedException, TimeoutException {
    // Construct the URI string identifying the machine type.
    String machineTypeUri = customMachineTypeUri(zone, cpuSeries, coreCount, memory);

    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests. After completing all of your requests, call
    // the `instancesClient.close()` method on the client to safely
    // clean up any remaining background resources.
    try (InstancesClient instancesClient = InstancesClient.create()) {

      AttachedDisk attachedDisk = AttachedDisk.newBuilder()
          .setInitializeParams(
              // Describe the size and source image of the boot disk to attach to the instance.
              // The list of public images available in Compute Engine can be found here:
              // https://cloud.google.com/compute/docs/images#list_of_public_images_available_on
              AttachedDiskInitializeParams.newBuilder()
                  .setSourceImage(
                      String.format("projects/%s/global/images/family/%s", "debian-cloud",
                          "debian-11"))
                  .setDiskSizeGb(10)
                  .build()
          )
          // Remember to set auto_delete to True if you want the disk to be deleted when you delete
          // your VM instance.
          .setAutoDelete(true)
          .setBoot(true)
          .build();

      // Create the Instance object with the relevant information.
      Instance instance = Instance.newBuilder()
          .setName(instanceName)
          .addDisks(attachedDisk)
          .setMachineType(machineTypeUri)
          .addNetworkInterfaces(
              NetworkInterface.newBuilder().setName("global/networks/default").build())
          .build();

      // Create the insert instance request object.
      InsertInstanceRequest insertInstanceRequest = InsertInstanceRequest.newBuilder()
          .setProject(project)
          .setZone(zone)
          .setInstanceResource(instance)
          .build();

      // Invoke the API with the request object and wait for the operation to complete.
      Operation response = instancesClient.insertAsync(insertInstanceRequest)
          .get(3, TimeUnit.MINUTES);

      // Check for errors.
      if (response.hasError()) {
        throw new Error("Instance creation failed!!" + response);
      }
      System.out.printf("Instance created : %s", instanceName);
      System.out.println("Operation Status: " + response.getStatus());
    }
  }

  public static String customMachineTypeUri(String zone, String cpuSeries, int coreCount,
      int memory) {

    if (!Arrays.asList(CpuSeries.E2.cpuSeries, CpuSeries.N1.cpuSeries, CpuSeries.N2.cpuSeries,
        CpuSeries.N2D.cpuSeries).contains(cpuSeries)) {
      throw new Error(String.format("Incorrect cpu type: %s", cpuSeries));
    }

    TypeLimits typeLimit = Objects.requireNonNull(
        typeLimitsMap.get(CpuSeries.get(cpuSeries).name())).typeLimits;

    // Perform the following checks to verify if the requested parameters are allowed.
    // Find more information about limitations of custom machine types at:
    // https://cloud.google.com/compute/docs/general-purpose-machines#custom_machine_types

    // 1. Check the number of cores and if the coreCount is present in allowedCores.
    if (typeLimit.allowedCores.length > 0 && Arrays.stream(typeLimit.allowedCores)
        .noneMatch(x -> x == coreCount)) {
      throw new Error(String.format(
          "Invalid number of cores requested. "
              + "Number of cores requested for CPU %s should be one of: %s",
          cpuSeries,
          Arrays.toString(typeLimit.allowedCores)));
    }

    // 2. Memory must be a multiple of 256 MB
    if (memory % 256 != 0) {
      throw new Error("Requested memory must be a multiple of 256 MB");
    }

    // 3. Check if the requested memory isn't too little
    if (memory < coreCount * typeLimit.minMemPerCore) {
      throw new Error(
          String.format("Requested memory is too low. Minimum memory for %s is %s MB per core",
              cpuSeries, typeLimit.minMemPerCore));
    }

    // 4. Check if the requested memory isn't too much
    if (memory > coreCount * typeLimit.maxMemPerCore && !typeLimit.allowExtraMemory) {
      throw new Error(String.format(
          "Requested memory is too large.. Maximum memory allowed for %s is %s MB per core",
          cpuSeries, typeLimit.extraMemoryLimit));
    }

    // 5. Check if the requested memory isn't too large
    if (memory > typeLimit.extraMemoryLimit && typeLimit.allowExtraMemory) {
      throw new Error(
          String.format("Requested memory is too large.. Maximum memory allowed for %s is %s MB",
              cpuSeries, typeLimit.extraMemoryLimit));
    }

    // Check if the CPU Series is E2 and return the custom machine type in the form of a string
    // acceptable by Compute Engine API.
    if (Arrays.asList(CpuSeries.E2_SMALL.cpuSeries, CpuSeries.E2_MICRO.cpuSeries,
        CpuSeries.E2_MEDIUM.cpuSeries).contains(cpuSeries)) {
      return String.format("zones/%s/machineTypes/%s-%s", zone, cpuSeries, memory);
    }

    // Check if extended memory was requested and return the extended custom machine type
    // in the form of a string acceptable by Compute Engine API.
    if (memory > coreCount * typeLimit.maxMemPerCore) {
      return String.format("zones/%s/machineTypes/%s-%s-%s-ext", zone, cpuSeries, coreCount,
          memory);
    }

    // Return the custom machine type in the form of a standard string
    // acceptable by Compute Engine API.
    return String.format("zones/%s/machineTypes/%s-%s-%s", zone, cpuSeries, coreCount, memory);
  }
}

Node.js

/**
 * TODO(developer): Uncomment and replace these variables before running the sample.
 */
// const projectId = 'YOUR_PROJECT_ID';
// const zone = 'europe-central2-b';
// const instanceName = 'YOUR_INSTANCE_NAME';
// const cpuSeries = 'N1';
// const coreCount = 2
// const memory = 256

const compute = require('@google-cloud/compute');

function range(from, to, step) {
  return [...Array(Math.floor((to - from) / step) + 1)].map(
    (_, i) => from + i * step
  );
}

class CustomMachineType {
  constructor(zone, cpuSeries, coreCount, memory) {
    this.zone = zone;
    this.cpuSeries = cpuSeries;
    this.coreCount = coreCount;
    this.memory = memory;

    this.N1 = 'custom';
    this.N2 = 'n2-custom';
    this.N2D = 'n2d-custom';
    this.E2 = 'e2-custom';
    this.E2Micro = 'e2-custom-micro';
    this.E2Small = 'e2-custom-small';
    this.E2Medium = 'e2-custom-medium';

    this.CpuSeriesE2Limit = {
      allowedCores: range(2, 33, 2),
      minMemPerCore: 512,
      maxMemPerCore: 8192,
      allowExtraMemory: false,
      extraMemoryLimit: 0,
    };

    this.CpuSeriesE2MicroLimit = {
      allowedCores: [],
      minMemPerCore: 1024,
      maxMemPerCore: 2048,
      allowExtraMemory: false,
      extraMemoryLimit: 0,
    };

    this.CpuSeriesE2SmallLimit = {
      allowedCores: [],
      minMemPerCore: 2048,
      maxMemPerCore: 4096,
      allowExtraMemory: false,
      extraMemoryLimit: 0,
    };

    this.CpuSeriesE2MediumLimit = {
      allowedCores: [],
      minMemPerCore: 4096,
      maxMemPerCore: 8192,
      allowExtraMemory: false,
      extraMemoryLimit: 0,
    };

    this.CpuSeriesN2Limit = {
      allowedCores: [...range(2, 33, 2), ...range(36, 129, 4)],
      minMemPerCore: 512,
      maxMemPerCore: 8192,
      allowExtraMemory: true,
      extraMemoryLimit: 624 << 10,
    };

    this.CpuSeriesN2DLimit = {
      allowedCores: [2, 4, 8, 16, 32, 48, 64, 80, 96],
      minMemPerCore: 512,
      maxMemPerCore: 8192,
      allowExtraMemory: true,
      extraMemoryLimit: 768 << 10,
    };

    this.CpuSeriesN1Limit = {
      allowedCores: [1, range(2, 97, 2)],
      minMemPerCore: 922,
      maxMemPerCore: 6656,
      allowExtraMemory: true,
      extraMemoryLimit: 624 << 10,
    };

    this.TYPE_LIMITS = {
      [this.N1]: this.CpuSeriesN1Limit,
      [this.N2]: this.CpuSeriesN2Limit,
      [this.N2D]: this.CpuSeriesN2DLimit,
      [this.E2]: this.CpuSeriesE2Limit,
      [this.E2Micro]: this.CpuSeriesE2MicroLimit,
      [this.E2Small]: this.CpuSeriesE2SmallLimit,
      [this.E2Medium]: this.CpuSeriesE2MediumLimit,
    };

    if (![this.E2, this.N1, this.N2, this.N2D].includes(cpuSeries)) {
      throw new Error(`Incorrect CPU type: ${this.cpuSeries}`);
    }

    this.typeLimit = this.TYPE_LIMITS[this.cpuSeries];

    // Check whether the requested parameters are allowed.
    // Find more information about limitations of custom machine types at:
    // https://cloud.google.com/compute/docs/general-purpose-machines#custom_machine_types

    // Check the number of cores
    if (
      this.typeLimit.allowedCores.length > 0 &&
      !this.typeLimit.allowedCores.includes(coreCount)
    ) {
      throw new Error(
        `Invalid number of cores requested. Allowed number of cores for ${this.cpuSeries} is: ${this.typeLimit.allowedCores}`
      );
    }

    // Memory must be a multiple of 256 MB
    if (this.memory % 256 !== 0) {
      throw new Error('Requested memory must be a multiple of 256 MB');
    }

    // Check if the requested memory isn't too little
    if (this.memory < this.coreCount * this.typeLimit.minMemPerCore) {
      throw new Error(
        `Requested memory is too low. Minimal memory for ${this.cpuSeries} is ${this.typeLimit.minMemPerCore} MB per core`
      );
    }

    // Check if the requested memory isn't too much
    if (
      this.memory > this.coreCount * this.typeLimit.maxMemPerCore &&
      !this.typeLimit.allowExtraMemory
    ) {
      throw new Error(
        `Requested memory is too large.. Maximum memory allowed for ${this.cpuSeries} is ${this.typeLimit.maxMemPerCore} MB per core`
      );
    }

    if (
      this.memory > this.typeLimit.extraMemoryLimit &&
      this.typeLimit.allowExtraMemory
    ) {
      throw new Error(
        `Requested memory is too large.. Maximum memory allowed for ${this.cpuSeries} is ${this.typeLimit.extraMemoryLimit} MB`
      );
    }
  }

  // Returns the custom machine type in form of a string acceptable by Compute Engine API.
  getMachineTypeURI() {
    if (
      [this.E2Small, this.E2Micro, this.E2Medium].includes(this.cpuSeries)
    ) {
      return `zones/${this.zone}/machineTypes/${this.cpuSeries}-${this.memory}`;
    }

    if (this.memory > this.coreCount * this.typeLimit.maxMemPerCore) {
      return `zones/${this.zone}/machineTypes/${this.cpuSeries}-${this.coreCount}-${this.memory}-ext`;
    }

    return `zones/${zone}/machineTypes/${this.cpuSeries}-${this.coreCount}-${this.memory}`;
  }
}

async function createInstanceWithCustomMachineTypeWithHelper() {
  const instancesClient = new compute.InstancesClient();

  const machineType = new CustomMachineType(
    zone,
    cpuSeries,
    coreCount,
    memory
  ).getMachineTypeURI();

  const [response] = await instancesClient.insert({
    instanceResource: {
      name: instanceName,
      disks: [
        {
          initializeParams: {
            diskSizeGb: '64',
            sourceImage:
              'projects/debian-cloud/global/images/family/debian-11/',
          },
          autoDelete: true,
          boot: true,
        },
      ],
      machineType,
      networkInterfaces: [
        {
          name: 'global/networks/default',
        },
      ],
    },
    project: projectId,
    zone,
  });
  let operation = response.latestResponse;
  const operationsClient = new compute.ZoneOperationsClient();

  // Wait for the create operation to complete.
  while (operation.status !== 'DONE') {
    [operation] = await operationsClient.wait({
      operation: operation.name,
      project: projectId,
      zone: operation.zone.split('/').pop(),
    });
  }

  console.log('Instance created.');
}

createInstanceWithCustomMachineTypeWithHelper();

Python

from collections import namedtuple
from enum import Enum
from enum import unique
import re
import sys
from typing import Any, List
import warnings

from google.api_core.extended_operation import ExtendedOperation
from google.cloud import compute_v1


def gb_to_mb(value: int) -> int:
    return value << 10


class CustomMachineType:
    """
    Allows to create custom machine types to be used with the VM instances.
    """

    @unique
    class CPUSeries(Enum):
        N1 = "custom"
        N2 = "n2-custom"
        N2D = "n2d-custom"
        E2 = "e2-custom"
        E2_MICRO = "e2-custom-micro"
        E2_SMALL = "e2-custom-small"
        E2_MEDIUM = "e2-custom-medium"

    TypeLimits = namedtuple(
        "TypeLimits",
        [
            "allowed_cores",
            "min_mem_per_core",
            "max_mem_per_core",
            "allow_extra_memory",
            "extra_memory_limit",
        ],
    )

    # The limits for various CPU types are described on:
    # https://cloud.google.com/compute/docs/general-purpose-machines
    LIMITS = {
        CPUSeries.E2: TypeLimits(frozenset(range(2, 33, 2)), 512, 8192, False, 0),
        CPUSeries.E2_MICRO: TypeLimits(frozenset(), 1024, 2048, False, 0),
        CPUSeries.E2_SMALL: TypeLimits(frozenset(), 2048, 4096, False, 0),
        CPUSeries.E2_MEDIUM: TypeLimits(frozenset(), 4096, 8192, False, 0),
        CPUSeries.N2: TypeLimits(
            frozenset(range(2, 33, 2)).union(set(range(36, 129, 4))),
            512,
            8192,
            True,
            gb_to_mb(624),
        ),
        CPUSeries.N2D: TypeLimits(
            frozenset({2, 4, 8, 16, 32, 48, 64, 80, 96}), 512, 8192, True, gb_to_mb(768)
        ),
        CPUSeries.N1: TypeLimits(
            frozenset({1}.union(range(2, 97, 2))), 922, 6656, True, gb_to_mb(624)
        ),
    }

    def __init__(
        self, zone: str, cpu_series: CPUSeries, memory_mb: int, core_count: int = 0
    ):
        self.zone = zone
        self.cpu_series = cpu_series
        self.limits = self.LIMITS[self.cpu_series]
        # Shared machine types (e2-small, e2-medium and e2-micro) always have
        # 2 vCPUs: https://cloud.google.com/compute/docs/general-purpose-machines#e2_limitations
        self.core_count = 2 if self.is_shared() else core_count
        self.memory_mb = memory_mb
        self._checked = False
        self._check_parameters()
        self.extra_memory_used = self._check_extra_memory()

    def is_shared(self):
        return self.cpu_series in (
            CustomMachineType.CPUSeries.E2_SMALL,
            CustomMachineType.CPUSeries.E2_MICRO,
            CustomMachineType.CPUSeries.E2_MEDIUM,
        )

    def _check_extra_memory(self) -> bool:
        if self._checked:
            return self.memory_mb > self.core_count * self.limits.max_mem_per_core
        else:
            raise RuntimeError(
                "You need to call _check_parameters() before calling _check_extra_memory()"
            )

    def _check_parameters(self):
        """
        Check whether the requested parameters are allowed. Find more information about limitations of custom machine
        types at: https://cloud.google.com/compute/docs/general-purpose-machines#custom_machine_types
        """
        # Check the number of cores
        if (
            self.limits.allowed_cores
            and self.core_count not in self.limits.allowed_cores
        ):
            raise RuntimeError(
                f"Invalid number of cores requested. Allowed number of cores for {self.cpu_series.name} is: {sorted(self.limits.allowed_cores)}"
            )

        # Memory must be a multiple of 256 MB
        if self.memory_mb % 256 != 0:
            raise RuntimeError("Requested memory must be a multiple of 256 MB.")

        # Check if the requested memory isn't too little
        if self.memory_mb < self.core_count * self.limits.min_mem_per_core:
            raise RuntimeError(
                f"Requested memory is too low. Minimal memory for {self.cpu_series.name} is {self.limits.min_mem_per_core} MB per core."
            )

        # Check if the requested memory isn't too much
        if self.memory_mb > self.core_count * self.limits.max_mem_per_core:
            if self.limits.allow_extra_memory:
                if self.memory_mb > self.limits.extra_memory_limit:
                    raise RuntimeError(
                        f"Requested memory is too large.. Maximum memory allowed for {self.cpu_series.name} is {self.limits.extra_memory_limit} MB."
                    )
            else:
                raise RuntimeError(
                    f"Requested memory is too large.. Maximum memory allowed for {self.cpu_series.name} is {self.limits.max_mem_per_core} MB per core."
                )

        self._checked = True

    def __str__(self) -> str:
        """
        Return the custom machine type in form of a string acceptable by Compute Engine API.
        """
        if self.cpu_series in {
            self.CPUSeries.E2_SMALL,
            self.CPUSeries.E2_MICRO,
            self.CPUSeries.E2_MEDIUM,
        }:
            return f"zones/{self.zone}/machineTypes/{self.cpu_series.value}-{self.memory_mb}"

        if self.extra_memory_used:
            return f"zones/{self.zone}/machineTypes/{self.cpu_series.value}-{self.core_count}-{self.memory_mb}-ext"

        return f"zones/{self.zone}/machineTypes/{self.cpu_series.value}-{self.core_count}-{self.memory_mb}"

    def short_type_str(self) -> str:
        """
        Return machine type in a format without the zone. For example, n2-custom-0-10240.
        This format is used to create instance templates.
        """
        return str(self).rsplit("/", maxsplit=1)[1]

    @classmethod
    def from_str(cls, machine_type: str):
        """
        Construct a new object from a string. The string needs to be a valid custom machine type like:
         - https://www.googleapis.com/compute/v1/projects/diregapic-mestiv/zones/us-central1-b/machineTypes/e2-custom-4-8192
         - zones/us-central1-b/machineTypes/e2-custom-4-8192
         - e2-custom-4-8192 (in this case, the zone parameter will not be set)
        """
        zone = None
        if machine_type.startswith("http"):
            machine_type = machine_type[machine_type.find("zones/") :]

        if machine_type.startswith("zones/"):
            _, zone, _, machine_type = machine_type.split("/")

        extra_mem = machine_type.endswith("-ext")

        if machine_type.startswith("custom"):
            cpu = cls.CPUSeries.N1
            _, cores, memory = machine_type.rsplit("-", maxsplit=2)
        else:
            if extra_mem:
                cpu_series, _, cores, memory, _ = machine_type.split("-")
            else:
                cpu_series, _, cores, memory = machine_type.split("-")
            if cpu_series == "n2":
                cpu = cls.CPUSeries.N2
            elif cpu_series == "n2d":
                cpu = cls.CPUSeries.N2D
            elif cpu_series == "e2":
                cpu = cls.CPUSeries.E2
                if cores == "micro":
                    cpu = cls.CPUSeries.E2_MICRO
                    cores = 2
                elif cores == "small":
                    cpu = cls.CPUSeries.E2_SMALL
                    cores = 2
                elif cores == "medium":
                    cpu = cls.CPUSeries.E2_MEDIUM
                    cores = 2
            else:
                raise RuntimeError("Unknown CPU series.")

        cores = int(cores)
        memory = int(memory)

        return cls(zone, cpu, memory, cores)


def get_image_from_family(project: str, family: str) -> compute_v1.Image:
    """
    Retrieve the newest image that is part of a given family in a project.

    Args:
        project: project ID or project number of the Cloud project you want to get image from.
        family: name of the image family you want to get image from.

    Returns:
        An Image object.
    """
    image_client = compute_v1.ImagesClient()
    # List of public operating system (OS) images: https://cloud.google.com/compute/docs/images/os-details
    newest_image = image_client.get_from_family(project=project, family=family)
    return newest_image


def disk_from_image(
    disk_type: str,
    disk_size_gb: int,
    boot: bool,
    source_image: str,
    auto_delete: bool = True,
) -> compute_v1.AttachedDisk:
    """
    Create an AttachedDisk object to be used in VM instance creation. Uses an image as the
    source for the new disk.

    Args:
         disk_type: the type of disk you want to create. This value uses the following format:
            "zones/{zone}/diskTypes/(pd-standard|pd-ssd|pd-balanced|pd-extreme)".
            For example: "zones/us-west3-b/diskTypes/pd-ssd"
        disk_size_gb: size of the new disk in gigabytes
        boot: boolean flag indicating whether this disk should be used as a boot disk of an instance
        source_image: source image to use when creating this disk. You must have read access to this disk. This can be one
            of the publicly available images or an image from one of your projects.
            This value uses the following format: "projects/{project_name}/global/images/{image_name}"
        auto_delete: boolean flag indicating whether this disk should be deleted with the VM that uses it

    Returns:
        AttachedDisk object configured to be created using the specified image.
    """
    boot_disk = compute_v1.AttachedDisk()
    initialize_params = compute_v1.AttachedDiskInitializeParams()
    initialize_params.source_image = source_image
    initialize_params.disk_size_gb = disk_size_gb
    initialize_params.disk_type = disk_type
    boot_disk.initialize_params = initialize_params
    # Remember to set auto_delete to True if you want the disk to be deleted when you delete
    # your VM instance.
    boot_disk.auto_delete = auto_delete
    boot_disk.boot = boot
    return boot_disk


def wait_for_extended_operation(
    operation: ExtendedOperation, verbose_name: str = "operation", timeout: int = 300
) -> Any:
    """
    This method will wait for the extended (long-running) operation to
    complete. If the operation is successful, it will return its result.
    If the operation ends with an error, an exception will be raised.
    If there were any warnings during the execution of the operation
    they will be printed to sys.stderr.

    Args:
        operation: a long-running operation you want to wait on.
        verbose_name: (optional) a more verbose name of the operation,
            used only during error and warning reporting.
        timeout: how long (in seconds) to wait for operation to finish.
            If None, wait indefinitely.

    Returns:
        Whatever the operation.result() returns.

    Raises:
        This method will raise the exception received from `operation.exception()`
        or RuntimeError if there is no exception set, but there is an `error_code`
        set for the `operation`.

        In case of an operation taking longer than `timeout` seconds to complete,
        a `concurrent.futures.TimeoutError` will be raised.
    """
    result = operation.result(timeout=timeout)

    if operation.error_code:
        print(
            f"Error during {verbose_name}: [Code: {operation.error_code}]: {operation.error_message}",
            file=sys.stderr,
            flush=True,
        )
        print(f"Operation ID: {operation.name}", file=sys.stderr, flush=True)
        raise operation.exception() or RuntimeError(operation.error_message)

    if operation.warnings:
        print(f"Warnings during {verbose_name}:\n", file=sys.stderr, flush=True)
        for warning in operation.warnings:
            print(f" - {warning.code}: {warning.message}", file=sys.stderr, flush=True)

    return result


def create_instance(
    project_id: str,
    zone: str,
    instance_name: str,
    disks: List[compute_v1.AttachedDisk],
    machine_type: str = "n1-standard-1",
    network_link: str = "global/networks/default",
    subnetwork_link: str = None,
    internal_ip: str = None,
    external_access: bool = False,
    external_ipv4: str = None,
    accelerators: List[compute_v1.AcceleratorConfig] = None,
    preemptible: bool = False,
    spot: bool = False,
    instance_termination_action: str = "STOP",
    custom_hostname: str = None,
    delete_protection: bool = False,
) -> compute_v1.Instance:
    """
    Send an instance creation request to the Compute Engine API and wait for it to complete.

    Args:
        project_id: project ID or project number of the Cloud project you want to use.
        zone: name of the zone to create the instance in. For example: "us-west3-b"
        instance_name: name of the new virtual machine (VM) instance.
        disks: a list of compute_v1.AttachedDisk objects describing the disks
            you want to attach to your new instance.
        machine_type: machine type of the VM being created. This value uses the
            following format: "zones/{zone}/machineTypes/{type_name}".
            For example: "zones/europe-west3-c/machineTypes/f1-micro"
        network_link: name of the network you want the new instance to use.
            For example: "global/networks/default" represents the network
            named "default", which is created automatically for each project.
        subnetwork_link: name of the subnetwork you want the new instance to use.
            This value uses the following format:
            "regions/{region}/subnetworks/{subnetwork_name}"
        internal_ip: internal IP address you want to assign to the new instance.
            By default, a free address from the pool of available internal IP addresses of
            used subnet will be used.
        external_access: boolean flag indicating if the instance should have an external IPv4
            address assigned.
        external_ipv4: external IPv4 address to be assigned to this instance. If you specify
            an external IP address, it must live in the same region as the zone of the instance.
            This setting requires `external_access` to be set to True to work.
        accelerators: a list of AcceleratorConfig objects describing the accelerators that will
            be attached to the new instance.
        preemptible: boolean value indicating if the new instance should be preemptible
            or not. Preemptible VMs have been deprecated and you should now use Spot VMs.
        spot: boolean value indicating if the new instance should be a Spot VM or not.
        instance_termination_action: What action should be taken once a Spot VM is terminated.
            Possible values: "STOP", "DELETE"
        custom_hostname: Custom hostname of the new VM instance.
            Custom hostnames must conform to RFC 1035 requirements for valid hostnames.
        delete_protection: boolean value indicating if the new virtual machine should be
            protected against deletion or not.
    Returns:
        Instance object.
    """
    instance_client = compute_v1.InstancesClient()

    # Use the network interface provided in the network_link argument.
    network_interface = compute_v1.NetworkInterface()
    network_interface.name = network_link
    if subnetwork_link:
        network_interface.subnetwork = subnetwork_link

    if internal_ip:
        network_interface.network_i_p = internal_ip

    if external_access:
        access = compute_v1.AccessConfig()
        access.type_ = compute_v1.AccessConfig.Type.ONE_TO_ONE_NAT.name
        access.name = "External NAT"
        access.network_tier = access.NetworkTier.PREMIUM.name
        if external_ipv4:
            access.nat_i_p = external_ipv4
        network_interface.access_configs = [access]

    # Collect information into the Instance object.
    instance = compute_v1.Instance()
    instance.network_interfaces = [network_interface]
    instance.name = instance_name
    instance.disks = disks
    if re.match(r"^zones/[a-z\d\-]+/machineTypes/[a-z\d\-]+$", machine_type):
        instance.machine_type = machine_type
    else:
        instance.machine_type = f"zones/{zone}/machineTypes/{machine_type}"

    if accelerators:
        instance.guest_accelerators = accelerators

    if preemptible:
        # Set the preemptible setting
        warnings.warn(
            "Preemptible VMs are being replaced by Spot VMs.", DeprecationWarning
        )
        instance.scheduling = compute_v1.Scheduling()
        instance.scheduling.preemptible = True

    if spot:
        # Set the Spot VM setting
        instance.scheduling = compute_v1.Scheduling()
        instance.scheduling.provisioning_model = (
            compute_v1.Scheduling.ProvisioningModel.SPOT.name
        )
        instance.scheduling.instance_termination_action = instance_termination_action

    if custom_hostname is not None:
        # Set the custom hostname for the instance
        instance.hostname = custom_hostname

    if delete_protection:
        # Set the delete protection bit
        instance.deletion_protection = True

    # Prepare the request to insert an instance.
    request = compute_v1.InsertInstanceRequest()
    request.zone = zone
    request.project = project_id
    request.instance_resource = instance

    # Wait for the create operation to complete.
    print(f"Creating the {instance_name} instance in {zone}...")

    operation = instance_client.insert(request=request)

    wait_for_extended_operation(operation, "instance creation")

    print(f"Instance {instance_name} created.")
    return instance_client.get(project=project_id, zone=zone, instance=instance_name)


def create_custom_instance(
    project_id: str,
    zone: str,
    instance_name: str,
    cpu_series: CustomMachineType.CPUSeries,
    core_count: int,
    memory: int,
) -> compute_v1.Instance:
    """
    Create a new VM instance with a custom machine type.

    Args:
        project_id: project ID or project number of the Cloud project you want to use.
        zone: name of the zone to create the instance in. For example: "us-west3-b"
        instance_name: name of the new virtual machine (VM) instance.
        cpu_series: the type of CPU you want to use. Select one value from the CustomMachineType.CPUSeries enum.
            For example: CustomMachineType.CPUSeries.N2
        core_count: number of CPU cores you want to use.
        memory: the amount of memory for the VM instance, in megabytes.

    Return:
        Instance object.
    """
    assert cpu_series in (
        CustomMachineType.CPUSeries.E2,
        CustomMachineType.CPUSeries.N1,
        CustomMachineType.CPUSeries.N2,
        CustomMachineType.CPUSeries.N2D,
    )
    custom_type = CustomMachineType(zone, cpu_series, memory, core_count)

    newest_debian = get_image_from_family(project="debian-cloud", family="debian-10")
    disk_type = f"zones/{zone}/diskTypes/pd-standard"
    disks = [disk_from_image(disk_type, 10, True, newest_debian.self_link)]

    return create_instance(project_id, zone, instance_name, disks, str(custom_type))

You can similarly create preemptible instances and groups of instances with custom machine types.

Add extended memory to a machine type

Depending on the machine, each machine type has a specific amount of memory by default. For example, if you create a custom N1 VM, it can have up to 6.5 GB of memory per vCPU. For custom N2 VMs, this number increases up to 8 GB of memory per vCPU.

For some workloads, this might not be enough memory. At an extra cost, you can get more memory per vCPU beyond the default limit. This is referred to as extended memory.

With extended memory, you can add memory to a machine type with no limitations per vCPU. You can add extended memory up to certain limits based on the machine type:

  • For N1 machine types, you can add up to a total of 624 GB of memory per VM.
  • For N2 machine types, you can add up to a total of 640 GB of memory per VM.
  • For N2D machine types, you can add up to a total of 768 GB of memory per VM.
  • Extended memory is not available for E2 machine types.

If you require more memory, you must use one of the mega-memory machine types, which lets you create instances with a total of 1.4 TB per VM instance.

Determining whether you need extended memory

Certain workloads need more than the default amount of memory per vCPU to deliver optimum results. Workloads that are based on in-memory high- performance analytics databases, including relational and NoSQL databases such as MS SQL Server, MongoDB, and MemcacheD/Redis, are in this category. The vCPU-based licensing of operating systems and software stacks also makes selecting an optimal VM memory configuration more challenging with predefined machine types. By using extended memory, you can add as much memory as your VMs need for the best price- performance ratio.

Pricing

When using a custom machine type, any memory up to and including the default amount per vCPU is charged at the N1 custom machine type pricing, or the N2 custom machine type pricing. E2 custom machine and E2 shared-core custom machine types share the same pricing scheme.

For memory above the default, Compute Engine charges for extended memory based on the N1 extended custom memory pricing or the N2 extended custom memory pricing. The price for extended memory is different from the price for memory below the default threshold.

Extended memory prices differ depending on the region. Instances running with extended memory are subject to the same 1-minute minimum charge as any other instance. Extended memory also qualifies for sustained use discounts.

For more information, see sustained use discounts for custom machine types and extended custom memory pricing.

Limitations

  • You can add extended memory only to custom machine types. Predefined machine types are not supported.
  • There is no limit to the memory per vCPU that you can add, but there is a maximum amount of memory you can add to each VM instance. For N1 machine types, you can add up to 624 GB of memory. For N2 machine types, you can add up to 640 GB of memory. For N2D machines types, you can add up to 768 GB of memory.
  • N2 and N2D machine types are available only in select regions and zones.
  • You must specify memory in 256-MB increments.
  • Extended memory does not qualify for committed use discounts.

Add extended memory during instance creation

To create a VM instance with extended memory, follow the instructions below.

Console

  1. In the Google Cloud console, go to the Create an instance page.

    Go to Create an instance

  2. In the Zone list, select the zone where you want to host this VM. The Series list is filtered to include only the machine type families available in the selected zone.

  3. In the Machine configuration section, select General-purpose.

    1. In the Series list, click N1 in First Generation for N1 custom machine types or N2 or N2D for Second Generation custom machine types.
    2. In the Machine type list, select Custom.
    3. To specify the number of vCPUs drag the Cores slider or enter the value in the text box.
    4. To add extended memory, select Extend memory, and do the following:

      1. To specify the amount of memory for the VM instance, drag the Memory slider or enter the value in the text box.

      The console displays an estimated cost for the VM as you change the number of vCPUs and memory.

  4. Continue to create the VM.

gcloud

For N1 machine types, use the gcloud command and the instances create sub-command, followed by the --custom-extensions flag to enable extended memory. Then, specify the total memory that you want with the --custom-memory parameter.

gcloud compute instances create example-instance  \
    --custom-extensions --custom-cpu=2 --custom-memory=15

For N2 machine types, use the gcloud command with the instances create sub-command. Set the --custom-extensions, --custom-cpu, and --custom-memory flags, followed by the --custom-vm-type flag, to specify the N2 machine type:

gcloud compute instances create example-instance \
    --custom-extensions --custom-cpu=2 --custom-memory=16 \
    --custom-vm-type=n2

For N2D machine types, use the gcloud command with the instances create sub-command. Set the --custom-extensions, --custom-cpu, and --custom-memory flags, followed by the --custom-vm-type flag, to specify the N2D machine type:

gcloud compute instances create example-instance \
    --custom-extensions --custom-cpu=2 --custom-memory=16 \
    --custom-vm-type=n2d

API

In the API, construct your instance creation request as you would normally, but when specifying a machineType value, use one of the following formats:

  • For N1 machine types, use:

    zones/ZONE/machineTypes/custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY-ext
    
  • For N2 machine types, use:

    zones/ZONE/machineTypes/n2-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY-ext
    
  • For N2D machine types, use:

    zones/ZONE/machineTypes/n2d-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY-ext
    

Replace the following:

  • NUMBER_OF_CPUS: The number of vCPUs.
  • AMOUNT_OF_MEMORY: The memory, in MB.

For example, the following specifies an N2 machine type with 2 vCPUs and 20 GB of memory:

zones/ZONE/machineTypes/n2-custom-2-20480-ext

Go

import (
	"context"
	"fmt"
	"io"

	compute "cloud.google.com/go/compute/apiv1"
	computepb "google.golang.org/genproto/googleapis/cloud/compute/v1"
	"google.golang.org/protobuf/proto"
)

// createInstanceWithExtraMemWithoutHelper сreates new VM instances with extra memory
// without using a CustomMachineType struct.
func createInstanceWithExtraMemWithoutHelper(
	w io.Writer,
	projectID, zone, instanceName, cpuSeries string,
	coreCount, memory int,
) error {
	// projectID := "your_project_id"
	// zone := "europe-central2-b"
	// instanceName := "your_instance_name"
	// cpuSeries := "N1"
	// coreCount := 2 // number of CPU cores you want to use.
	// memory := 256 // the amount of memory for the VM instance, in megabytes.

	// The coreCount and memory values are not validated anywhere and can be rejected by the API.

	ctx := context.Background()
	instancesClient, err := compute.NewInstancesRESTClient(ctx)
	if err != nil {
		return fmt.Errorf("NewInstancesRESTClient: %v", err)
	}
	defer instancesClient.Close()

	mt := fmt.Sprintf("zones/%s/machineTypes/%v-%v-%v-ext", zone, cpuSeries, coreCount, memory)
	inst := &computepb.Instance{
		Name: proto.String(instanceName),
		Disks: []*computepb.AttachedDisk{
			{
				InitializeParams: &computepb.AttachedDiskInitializeParams{
					DiskSizeGb: proto.Int64(10),
					SourceImage: proto.String(
						"projects/debian-cloud/global/images/family/debian-10",
					),
				},
				AutoDelete: proto.Bool(true),
				Boot:       proto.Bool(true),
			},
		},
		MachineType: proto.String(mt),
		NetworkInterfaces: []*computepb.NetworkInterface{
			{
				Name: proto.String("global/networks/default"),
			},
		},
	}

	req := &computepb.InsertInstanceRequest{
		Project:          projectID,
		Zone:             zone,
		InstanceResource: inst,
	}

	op, err := instancesClient.Insert(ctx, req)
	if err != nil {
		return fmt.Errorf("unable to create instance: %v", err)
	}

	if err = op.Wait(ctx); err != nil {
		return fmt.Errorf("unable to wait for the operation: %v", err)
	}

	fmt.Fprintf(w, "Instance created\n")

	return nil
}

Java


import com.google.cloud.compute.v1.AttachedDisk;
import com.google.cloud.compute.v1.AttachedDiskInitializeParams;
import com.google.cloud.compute.v1.InsertInstanceRequest;
import com.google.cloud.compute.v1.Instance;
import com.google.cloud.compute.v1.InstancesClient;
import com.google.cloud.compute.v1.NetworkInterface;
import com.google.cloud.compute.v1.Operation;
import java.io.IOException;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;

public class ExtraMemoryWithoutHelper {

  public static void main(String[] args)
      throws IOException, ExecutionException, InterruptedException, TimeoutException {
    // TODO(developer): Replace these variables before running the sample.
    // Project ID or project number of the Cloud project you want to use.
    String projectId = "your-google-cloud-project-id";
    // Name of the zone to create the instance in. For example: "us-west3-b".
    String zone = "google-cloud-zone";
    // Name of the new virtual machine (VM) instance.
    String instanceName = "instance-name";
    String cpuSeries = "N1";
    // Number of CPU cores you want to use.
    int coreCount = 2;
    // The amount of memory for the VM instance, in megabytes.
    int memory = 256;

    createInstanceWithExtraMemoryWithoutHelper(projectId, zone, instanceName, cpuSeries, coreCount,
        memory);
  }

  // Create VM instances with extra memory without using a CustomMachineType class and
  // return the created Instance.
  public static void createInstanceWithExtraMemoryWithoutHelper(
      String project, String zone, String instanceName, String cpuSeries, int coreCount, int memory)
      throws IOException, ExecutionException, InterruptedException, TimeoutException {
    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests. After completing all of your requests, call
    // the `instancesClient.close()` method on the client to safely
    // clean up any remaining background resources.
    try (InstancesClient instancesClient = InstancesClient.create()) {

      // The coreCount and memory values are not validated anywhere and can be rejected by the API.
      String machineType = String.format("zones/%s/machineTypes/%s-%s-%s-ext", zone, cpuSeries,
          coreCount, memory);

      AttachedDisk attachedDisk = AttachedDisk.newBuilder()
          .setInitializeParams(
              // Describe the size and source image of the boot disk to attach to the instance.
              // The list of public images available in Compute Engine can be found here:
              // https://cloud.google.com/compute/docs/images#list_of_public_images_available_on
              AttachedDiskInitializeParams.newBuilder()
                  .setSourceImage(
                      String.format("projects/%s/global/images/family/%s", "debian-cloud",
                          "debian-11"))
                  .setDiskSizeGb(10)
                  .build()
          )
          // Remember to set auto_delete to True if you want the disk to be deleted when you delete
          // your VM instance.
          .setAutoDelete(true)
          .setBoot(true)
          .build();

      // Create the Instance object with the relevant information.
      Instance instance = Instance.newBuilder()
          .setName(instanceName)
          .addDisks(attachedDisk)
          .setMachineType(machineType)
          .addNetworkInterfaces(
              NetworkInterface.newBuilder().setName("global/networks/default").build())
          .build();

      // Create the insert instance request object.
      InsertInstanceRequest insertInstanceRequest = InsertInstanceRequest.newBuilder()
          .setProject(project)
          .setZone(zone)
          .setInstanceResource(instance)
          .build();

      // Invoke the API with the request object and wait for the operation to complete.
      Operation response = instancesClient.insertAsync(insertInstanceRequest)
          .get(3, TimeUnit.MINUTES);

      // Check for errors.
      if (response.hasError()) {
        System.out.println("Instance creation failed!!" + response);
        return;
      }
      System.out.printf("Instance created : %s", instanceName);
      System.out.println("Operation Status: " + response.getStatus());
    }
  }
}

Node.js

/**
 * TODO(developer): Uncomment and replace these variables before running the sample.
 */
// const projectId = 'YOUR_PROJECT_ID';
// const zone = 'europe-central2-b';
// const instanceName = 'YOUR_INSTANCE_NAME';
// const cpuSeries = 'N1';
// const coreCount = 2
// const memory = 256

// The coreCount and memory values are not validated anywhere and can be rejected by the API.

const compute = require('@google-cloud/compute');

async function createInstanceWithExtraMemWithoutHelper() {
  const instancesClient = new compute.InstancesClient();

  const machineType = `zones/${zone}/machineTypes/${cpuSeries}-${coreCount}-${memory}-ext`;

  const [response] = await instancesClient.insert({
    instanceResource: {
      name: instanceName,
      disks: [
        {
          initializeParams: {
            diskSizeGb: '64',
            sourceImage:
              'projects/debian-cloud/global/images/family/debian-11/',
          },
          autoDelete: true,
          boot: true,
        },
      ],
      machineType,
      networkInterfaces: [
        {
          name: 'global/networks/default',
        },
      ],
    },
    project: projectId,
    zone,
  });
  let operation = response.latestResponse;
  const operationsClient = new compute.ZoneOperationsClient();

  // Wait for the create operation to complete.
  while (operation.status !== 'DONE') {
    [operation] = await operationsClient.wait({
      operation: operation.name,
      project: projectId,
      zone: operation.zone.split('/').pop(),
    });
  }

  console.log('Instance created.');
}

createInstanceWithExtraMemWithoutHelper();

Python

import re
import sys
from typing import Any, List
import warnings

from google.api_core.extended_operation import ExtendedOperation
from google.cloud import compute_v1


def get_image_from_family(project: str, family: str) -> compute_v1.Image:
    """
    Retrieve the newest image that is part of a given family in a project.

    Args:
        project: project ID or project number of the Cloud project you want to get image from.
        family: name of the image family you want to get image from.

    Returns:
        An Image object.
    """
    image_client = compute_v1.ImagesClient()
    # List of public operating system (OS) images: https://cloud.google.com/compute/docs/images/os-details
    newest_image = image_client.get_from_family(project=project, family=family)
    return newest_image


def disk_from_image(
    disk_type: str,
    disk_size_gb: int,
    boot: bool,
    source_image: str,
    auto_delete: bool = True,
) -> compute_v1.AttachedDisk:
    """
    Create an AttachedDisk object to be used in VM instance creation. Uses an image as the
    source for the new disk.

    Args:
         disk_type: the type of disk you want to create. This value uses the following format:
            "zones/{zone}/diskTypes/(pd-standard|pd-ssd|pd-balanced|pd-extreme)".
            For example: "zones/us-west3-b/diskTypes/pd-ssd"
        disk_size_gb: size of the new disk in gigabytes
        boot: boolean flag indicating whether this disk should be used as a boot disk of an instance
        source_image: source image to use when creating this disk. You must have read access to this disk. This can be one
            of the publicly available images or an image from one of your projects.
            This value uses the following format: "projects/{project_name}/global/images/{image_name}"
        auto_delete: boolean flag indicating whether this disk should be deleted with the VM that uses it

    Returns:
        AttachedDisk object configured to be created using the specified image.
    """
    boot_disk = compute_v1.AttachedDisk()
    initialize_params = compute_v1.AttachedDiskInitializeParams()
    initialize_params.source_image = source_image
    initialize_params.disk_size_gb = disk_size_gb
    initialize_params.disk_type = disk_type
    boot_disk.initialize_params = initialize_params
    # Remember to set auto_delete to True if you want the disk to be deleted when you delete
    # your VM instance.
    boot_disk.auto_delete = auto_delete
    boot_disk.boot = boot
    return boot_disk


def wait_for_extended_operation(
    operation: ExtendedOperation, verbose_name: str = "operation", timeout: int = 300
) -> Any:
    """
    This method will wait for the extended (long-running) operation to
    complete. If the operation is successful, it will return its result.
    If the operation ends with an error, an exception will be raised.
    If there were any warnings during the execution of the operation
    they will be printed to sys.stderr.

    Args:
        operation: a long-running operation you want to wait on.
        verbose_name: (optional) a more verbose name of the operation,
            used only during error and warning reporting.
        timeout: how long (in seconds) to wait for operation to finish.
            If None, wait indefinitely.

    Returns:
        Whatever the operation.result() returns.

    Raises:
        This method will raise the exception received from `operation.exception()`
        or RuntimeError if there is no exception set, but there is an `error_code`
        set for the `operation`.

        In case of an operation taking longer than `timeout` seconds to complete,
        a `concurrent.futures.TimeoutError` will be raised.
    """
    result = operation.result(timeout=timeout)

    if operation.error_code:
        print(
            f"Error during {verbose_name}: [Code: {operation.error_code}]: {operation.error_message}",
            file=sys.stderr,
            flush=True,
        )
        print(f"Operation ID: {operation.name}", file=sys.stderr, flush=True)
        raise operation.exception() or RuntimeError(operation.error_message)

    if operation.warnings:
        print(f"Warnings during {verbose_name}:\n", file=sys.stderr, flush=True)
        for warning in operation.warnings:
            print(f" - {warning.code}: {warning.message}", file=sys.stderr, flush=True)

    return result


def create_instance(
    project_id: str,
    zone: str,
    instance_name: str,
    disks: List[compute_v1.AttachedDisk],
    machine_type: str = "n1-standard-1",
    network_link: str = "global/networks/default",
    subnetwork_link: str = None,
    internal_ip: str = None,
    external_access: bool = False,
    external_ipv4: str = None,
    accelerators: List[compute_v1.AcceleratorConfig] = None,
    preemptible: bool = False,
    spot: bool = False,
    instance_termination_action: str = "STOP",
    custom_hostname: str = None,
    delete_protection: bool = False,
) -> compute_v1.Instance:
    """
    Send an instance creation request to the Compute Engine API and wait for it to complete.

    Args:
        project_id: project ID or project number of the Cloud project you want to use.
        zone: name of the zone to create the instance in. For example: "us-west3-b"
        instance_name: name of the new virtual machine (VM) instance.
        disks: a list of compute_v1.AttachedDisk objects describing the disks
            you want to attach to your new instance.
        machine_type: machine type of the VM being created. This value uses the
            following format: "zones/{zone}/machineTypes/{type_name}".
            For example: "zones/europe-west3-c/machineTypes/f1-micro"
        network_link: name of the network you want the new instance to use.
            For example: "global/networks/default" represents the network
            named "default", which is created automatically for each project.
        subnetwork_link: name of the subnetwork you want the new instance to use.
            This value uses the following format:
            "regions/{region}/subnetworks/{subnetwork_name}"
        internal_ip: internal IP address you want to assign to the new instance.
            By default, a free address from the pool of available internal IP addresses of
            used subnet will be used.
        external_access: boolean flag indicating if the instance should have an external IPv4
            address assigned.
        external_ipv4: external IPv4 address to be assigned to this instance. If you specify
            an external IP address, it must live in the same region as the zone of the instance.
            This setting requires `external_access` to be set to True to work.
        accelerators: a list of AcceleratorConfig objects describing the accelerators that will
            be attached to the new instance.
        preemptible: boolean value indicating if the new instance should be preemptible
            or not. Preemptible VMs have been deprecated and you should now use Spot VMs.
        spot: boolean value indicating if the new instance should be a Spot VM or not.
        instance_termination_action: What action should be taken once a Spot VM is terminated.
            Possible values: "STOP", "DELETE"
        custom_hostname: Custom hostname of the new VM instance.
            Custom hostnames must conform to RFC 1035 requirements for valid hostnames.
        delete_protection: boolean value indicating if the new virtual machine should be
            protected against deletion or not.
    Returns:
        Instance object.
    """
    instance_client = compute_v1.InstancesClient()

    # Use the network interface provided in the network_link argument.
    network_interface = compute_v1.NetworkInterface()
    network_interface.name = network_link
    if subnetwork_link:
        network_interface.subnetwork = subnetwork_link

    if internal_ip:
        network_interface.network_i_p = internal_ip

    if external_access:
        access = compute_v1.AccessConfig()
        access.type_ = compute_v1.AccessConfig.Type.ONE_TO_ONE_NAT.name
        access.name = "External NAT"
        access.network_tier = access.NetworkTier.PREMIUM.name
        if external_ipv4:
            access.nat_i_p = external_ipv4
        network_interface.access_configs = [access]

    # Collect information into the Instance object.
    instance = compute_v1.Instance()
    instance.network_interfaces = [network_interface]
    instance.name = instance_name
    instance.disks = disks
    if re.match(r"^zones/[a-z\d\-]+/machineTypes/[a-z\d\-]+$", machine_type):
        instance.machine_type = machine_type
    else:
        instance.machine_type = f"zones/{zone}/machineTypes/{machine_type}"

    if accelerators:
        instance.guest_accelerators = accelerators

    if preemptible:
        # Set the preemptible setting
        warnings.warn(
            "Preemptible VMs are being replaced by Spot VMs.", DeprecationWarning
        )
        instance.scheduling = compute_v1.Scheduling()
        instance.scheduling.preemptible = True

    if spot:
        # Set the Spot VM setting
        instance.scheduling = compute_v1.Scheduling()
        instance.scheduling.provisioning_model = (
            compute_v1.Scheduling.ProvisioningModel.SPOT.name
        )
        instance.scheduling.instance_termination_action = instance_termination_action

    if custom_hostname is not None:
        # Set the custom hostname for the instance
        instance.hostname = custom_hostname

    if delete_protection:
        # Set the delete protection bit
        instance.deletion_protection = True

    # Prepare the request to insert an instance.
    request = compute_v1.InsertInstanceRequest()
    request.zone = zone
    request.project = project_id
    request.instance_resource = instance

    # Wait for the create operation to complete.
    print(f"Creating the {instance_name} instance in {zone}...")

    operation = instance_client.insert(request=request)

    wait_for_extended_operation(operation, "instance creation")

    print(f"Instance {instance_name} created.")
    return instance_client.get(project=project_id, zone=zone, instance=instance_name)


def create_custom_instances_extra_mem(
    project_id: str, zone: str, instance_name: str, core_count: int, memory: int
) -> List[compute_v1.Instance]:
    """
    Create 3 new VM instances with extra memory without using a CustomMachineType helper class.

    Args:
        project_id: project ID or project number of the Cloud project you want to use.
        zone: name of the zone to create the instance in. For example: "us-west3-b"
        instance_name: name of the new virtual machine (VM) instance.
        core_count: number of CPU cores you want to use.
        memory: the amount of memory for the VM instance, in megabytes.

    Returns:
        List of Instance objects.
    """
    newest_debian = get_image_from_family(project="debian-cloud", family="debian-10")
    disk_type = f"zones/{zone}/diskTypes/pd-standard"
    disks = [disk_from_image(disk_type, 10, True, newest_debian.self_link)]
    # The core_count and memory values are not validated anywhere and can be rejected by the API.
    instances = [
        create_instance(
            project_id,
            zone,
            f"{instance_name}_n1_extra_mem",
            disks,
            f"zones/{zone}/machineTypes/custom-{core_count}-{memory}-ext",
        ),
        create_instance(
            project_id,
            zone,
            f"{instance_name}_n2_extra_mem",
            disks,
            f"zones/{zone}/machineTypes/n2-custom-{core_count}-{memory}-ext",
        ),
        create_instance(
            project_id,
            zone,
            f"{instance_name}_n2d_extra_mem",
            disks,
            f"zones/{zone}/machineTypes/n2d-custom-{core_count}-{memory}-ext",
        ),
    ]
    return instances

Add extended memory to an existing VM instance

To add more memory to an existing instance, you must first stop the instance. After the instance stops, follow the instructions below to add more memory to the VM.

Console

  1. In the Google Cloud console, go to the VM instances page.

    Go to VM instances

  2. Choose the stopped VM you want to modify from the VM list.

  3. Click Edit at the top of the page.

  4. In Machine configuration, select General-purpose.

  5. From the Machine type list, select Custom.

  6. Select the number of vCPUs you want.

  7. To add extended memory, select Extend memory and then specify the amount of memory you want.

  8. Save your changes.

gcloud

Using the Google Cloud CLI, use the instances set-machine-type sub-command, followed by the --custom-extensions flag paired with the existing --custom-memory parameter.

For example, to add 2 GB memory to a VM instance named example-instance, which currently has an N1 machine type with 2 vCPU, 13 GB, in the us-central1-b zone, use the following command:

gcloud compute instances set-machine-type example-instance \
   --zone=us-central1-b --custom-cpu=2 \
   --custom-memory=15 --custom-extensions

API

In the API, make a POST request to the setMachineType method with the desired machine type in the request body. Use one of the following formats:

  • For N1 machine types, use:

    zones/ZONE/machineTypes/custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY-ext
    
  • For N2 machine types, use:

    zones/ZONE/machineTypes/n2-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY-ext
    
  • For N2D machine types, use:

    zones/ZONE/machineTypes/n2d-custom-NUMBER_OF_CPUS-AMOUNT_OF_MEMORY-ext
    

Replace the following:

  • NUMBER_OF_CPUS: The number of vCPUs.
  • AMOUNT_OF_MEMORY: The memory, in MB.

For example, the following request uses a machine type with 2 vCPUs and 15 GB of memory:

zones/ZONE/machineTypes/custom-2-15360-ext

For example:

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/instances/INSTANCE_NAME/setMachineType

{
    machineType: "zones/us-central1-f/machineTypes/custom-2-15360-ext"
}

Go

import (
	"context"
	"fmt"
	"io"
	"strings"

	compute "cloud.google.com/go/compute/apiv1"
	computepb "google.golang.org/genproto/googleapis/cloud/compute/v1"
	"google.golang.org/protobuf/proto"
)

// modifyInstanceWithExtendedMemory sends an instance creation request
// to the Compute Engine API and waits for it to complete.
func modifyInstanceWithExtendedMemory(
	w io.Writer,
	projectID, zone, instanceName string,
	newMemory int,
) error {
	// projectID := "your_project_id"
	// zone := "europe-central2-b"
	// instanceName := "your_instance_name"
	// newMemory := 256 // the amount of memory for the VM instance, in megabytes.

	ctx := context.Background()
	instancesClient, err := compute.NewInstancesRESTClient(ctx)
	if err != nil {
		return fmt.Errorf("NewInstancesRESTClient: %v", err)
	}
	defer instancesClient.Close()

	reqInstance := &computepb.GetInstanceRequest{
		Project:  projectID,
		Zone:     zone,
		Instance: instanceName,
	}

	instance, err := instancesClient.Get(ctx, reqInstance)
	if err != nil {
		return fmt.Errorf("unable to get instance: %v", err)
	}

	containsString := func(s []string, str string) bool {
		for _, v := range s {
			if v == str {
				return true
			}
		}

		return false
	}

	if !(strings.Contains(instance.GetMachineType(), "machineTypes/n1-") ||
		strings.Contains(instance.GetMachineType(), "machineTypes/n2-") ||
		strings.Contains(instance.GetMachineType(), "machineTypes/n2d-")) {
		return fmt.Errorf("extra memory is available only for N1, N2 and N2D CPUs")
	}

	// Make sure that the machine is turned off
	if !containsString([]string{"TERMINATED", "STOPPED"}, instance.GetStatus()) {
		reqStop := &computepb.StopInstanceRequest{
			Project:  projectID,
			Zone:     zone,
			Instance: instanceName,
		}

		op, err := instancesClient.Stop(ctx, reqStop)
		if err != nil {
			return fmt.Errorf("unable to stop instance: %v", err)
		}

		if err = op.Wait(ctx); err != nil {
			return fmt.Errorf("unable to wait for the operation: %v", err)
		}
	}

	// Modify the machine definition, remember that extended memory
	// is available only for N1, N2 and N2D CPUs
	machineType := instance.GetMachineType()
	start := machineType[:strings.LastIndex(machineType, "-")]

	updateReq := &computepb.SetMachineTypeInstanceRequest{
		Project:  projectID,
		Zone:     zone,
		Instance: instanceName,
		InstancesSetMachineTypeRequestResource: &computepb.InstancesSetMachineTypeRequest{
			MachineType: proto.String(fmt.Sprintf("%s-%v-ext", start, newMemory)),
		},
	}
	op, err := instancesClient.SetMachineType(ctx, updateReq)
	if err != nil {
		return fmt.Errorf("unable to update instance: %v", err)
	}

	if err = op.Wait(ctx); err != nil {
		return fmt.Errorf("unable to wait for the operation: %v", err)
	}

	fmt.Fprintf(w, "Instance updated\n")

	return nil

}

Java


import com.google.api.gax.longrunning.OperationFuture;
import com.google.cloud.compute.v1.GetInstanceRequest;
import com.google.cloud.compute.v1.Instance;
import com.google.cloud.compute.v1.Instance.Status;
import com.google.cloud.compute.v1.InstancesClient;
import com.google.cloud.compute.v1.InstancesSetMachineTypeRequest;
import com.google.cloud.compute.v1.Operation;
import com.google.cloud.compute.v1.SetMachineTypeInstanceRequest;
import com.google.cloud.compute.v1.StopInstanceRequest;
import java.io.IOException;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;

public class UpdateMemory {

  public static void main(String[] args)
      throws IOException, ExecutionException, InterruptedException, TimeoutException {
    // TODO(developer): Replace these variables before running the sample.
    // Project ID or project number of the Cloud project you want to use.
    String projectId = "your-google-cloud-project-id";
    // Name of the zone to create the instance in. For example: "us-west3-b".
    String zone = "google-cloud-zone";
    // Name of the new virtual machine (VM) instance.
    String instanceName = "instance-name";
    // The amount of memory for the VM instance, in megabytes.
    int newMemory = 256;

    modifyInstanceWithExtendedMemory(projectId, zone, instanceName, newMemory);
  }

  // Modify an existing VM to use extended memory and return the modified Instance.
  public static void modifyInstanceWithExtendedMemory(
      String project, String zone, String instanceName, int newMemory)
      throws IOException, ExecutionException, InterruptedException, TimeoutException {
    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests. After completing all of your requests, call
    // the `instancesClient.close()` method on the client to safely
    // clean up any remaining background resources.
    try (InstancesClient instancesClient = InstancesClient.create()) {

      // Create the get instance request object.
      GetInstanceRequest getInstanceRequest = GetInstanceRequest.newBuilder()
          .setProject(project)
          .setZone(zone)
          .setInstance(instanceName)
          .build();

      Instance instance = instancesClient.get(getInstanceRequest);

      // Check the machine type.
      if (!(instance.getMachineType().contains("machineTypes/n1-")
          || instance.getMachineType().contains("machineTypes/n2-")
          || instance.getMachineType().contains("machineTypes/n2d-"))) {
        System.out.println("extra memory is available only for N1, N2 and N2D CPUs");
        return;
      }

      // Make sure that the machine is turned off.
      if (!(instance.getStatus().equals(Status.TERMINATED.toString())
          || instance.getStatus().equals(Status.STOPPED.toString()))) {

        StopInstanceRequest stopInstanceRequest = StopInstanceRequest.newBuilder()
            .setProject(project)
            .setZone(zone)
            .setInstance(instanceName)
            .build();

        OperationFuture<Operation, Operation> operation = instancesClient.stopAsync(
            stopInstanceRequest);
        Operation response = operation.get(3, TimeUnit.MINUTES);
        if (response.hasError()) {
          System.out.printf("Unable to stop instance %s", response.getError());
          return;
        }
      }

      // Modify the machine definition. Note that extended memory
      // is available only for N1, N2 and N2D CPUs.
      String machineType = instance.getMachineType();
      String start = machineType.substring(0, machineType.lastIndexOf("-"));

      // Create the machine type instance request object.
      SetMachineTypeInstanceRequest setMachineTypeInstanceRequest =
          SetMachineTypeInstanceRequest.newBuilder()
              .setProject(project)
              .setZone(zone)
              .setInstance(instanceName)
              .setInstancesSetMachineTypeRequestResource(InstancesSetMachineTypeRequest.newBuilder()
                  .setMachineType(String.format("%s-%d-ext", start, newMemory))
                  .build())
              .build();

      // Invoke the API with the request object and wait for the operation to complete.
      Operation response = instancesClient.setMachineTypeAsync(setMachineTypeInstanceRequest)
          .get(3, TimeUnit.MINUTES);

      // Check for errors.
      if (response.hasError()) {
        System.out.printf("Unable to update instance %s", response.getError());
        return;
      }
      System.out.println("Instance updated!");
    }
  }
}

Node.js

/**
 * TODO(developer): Uncomment and replace these variables before running the sample.
 */
// const projectId = 'YOUR_PROJECT_ID';
// const zone = 'europe-central2-b';
// const instanceName = 'YOUR_INSTANCE_NAME';
// const newMemory = 256;

const compute = require('@google-cloud/compute');

async function modifyInstanceWithExtendedMemory() {
  const instancesClient = new compute.InstancesClient();

  const [instance] = await instancesClient.get({
    project: projectId,
    zone,
    instance: instanceName,
  });

  if (
    !['machineTypes/n1-', 'machineTypes/n2-', 'machineTypes/n2d-'].some(
      type => instance.machineType.includes(type)
    )
  ) {
    throw new Error('extra memory is available only for N1, N2 and N2D CPUs');
  }

  // Make sure that the machine is turned off
  if (!['TERMINATED', 'STOPPED'].some(status => instance.status === status)) {
    const [response] = await instancesClient.stop({
      project: projectId,
      zone,
      instance: instanceName,
    });

    let operation = response.latestResponse;
    const operationsClient = new compute.ZoneOperationsClient();

    // Wait for the stop operation to complete.
    while (operation.status !== 'DONE') {
      [operation] = await operationsClient.wait({
        operation: operation.name,
        project: projectId,
        zone: operation.zone.split('/').pop(),
      });
    }
  }

  // Modify the machine definition, remember that extended memory
  // is available only for N1, N2 and N2D CPUs

  const start = instance.machineType.substring(
    0,
    instance.machineType.lastIndexOf('-')
  );

  const [response] = await instancesClient.setMachineType({
    project: projectId,
    zone,
    instance: instanceName,
    instancesSetMachineTypeRequestResource: {
      machineType: `${start}-${newMemory}-ext`,
    },
  });
  let operation = response.latestResponse;
  const operationsClient = new compute.ZoneOperationsClient();

  // Wait for the update operation to complete.
  while (operation.status !== 'DONE') {
    [operation] = await operationsClient.wait({
      operation: operation.name,
      project: projectId,
      zone: operation.zone.split('/').pop(),
    });
  }

  console.log('Instance updated.');
}

modifyInstanceWithExtendedMemory();

Python

import sys
import time
from typing import Any

from google.api_core.extended_operation import ExtendedOperation
from google.cloud import compute_v1


def wait_for_extended_operation(
    operation: ExtendedOperation, verbose_name: str = "operation", timeout: int = 300
) -> Any:
    """
    This method will wait for the extended (long-running) operation to
    complete. If the operation is successful, it will return its result.
    If the operation ends with an error, an exception will be raised.
    If there were any warnings during the execution of the operation
    they will be printed to sys.stderr.

    Args:
        operation: a long-running operation you want to wait on.
        verbose_name: (optional) a more verbose name of the operation,
            used only during error and warning reporting.
        timeout: how long (in seconds) to wait for operation to finish.
            If None, wait indefinitely.

    Returns:
        Whatever the operation.result() returns.

    Raises:
        This method will raise the exception received from `operation.exception()`
        or RuntimeError if there is no exception set, but there is an `error_code`
        set for the `operation`.

        In case of an operation taking longer than `timeout` seconds to complete,
        a `concurrent.futures.TimeoutError` will be raised.
    """
    result = operation.result(timeout=timeout)

    if operation.error_code:
        print(
            f"Error during {verbose_name}: [Code: {operation.error_code}]: {operation.error_message}",
            file=sys.stderr,
            flush=True,
        )
        print(f"Operation ID: {operation.name}", file=sys.stderr, flush=True)
        raise operation.exception() or RuntimeError(operation.error_message)

    if operation.warnings:
        print(f"Warnings during {verbose_name}:\n", file=sys.stderr, flush=True)
        for warning in operation.warnings:
            print(f" - {warning.code}: {warning.message}", file=sys.stderr, flush=True)

    return result


def add_extended_memory_to_instance(
    project_id: str, zone: str, instance_name: str, new_memory: int
):
    """
    Modify an existing VM to use extended memory.

    Args:
        project_id: project ID or project number of the Cloud project you want to use.
        zone: name of the zone to create the instance in. For example: "us-west3-b"
        instance_name: name of the new virtual machine (VM) instance.
        new_memory: the amount of memory for the VM instance, in megabytes.

    Returns:
        Instance object.
    """
    instance_client = compute_v1.InstancesClient()
    instance = instance_client.get(
        project=project_id, zone=zone, instance=instance_name
    )

    if not (
        "n1-" in instance.machine_type
        or "n2-" in instance.machine_type
        or "n2d-" in instance.machine_type
    ):
        raise RuntimeError("Extra memory is available only for N1, N2 and N2D CPUs.")

    # Make sure that the machine is turned off
    if instance.status not in (
        instance.Status.TERMINATED.name,
        instance.Status.STOPPED.name,
    ):
        operation = instance_client.stop(
            project=project_id, zone=zone, instance=instance_name
        )
        wait_for_extended_operation(operation, "instance stopping")
        start = time.time()
        while instance.status not in (
            instance.Status.TERMINATED.name,
            instance.Status.STOPPED.name,
        ):
            # Waiting for the instance to be turned off.
            instance = instance_client.get(
                project=project_id, zone=zone, instance=instance_name
            )
            time.sleep(2)
            if time.time() - start >= 300:  # 5 minutes
                raise TimeoutError()

    # Modify the machine definition, remember that extended memory is available only for N1, N2 and N2D CPUs
    start, end = instance.machine_type.rsplit("-", maxsplit=1)
    instance.machine_type = start + f"-{new_memory}-ext"
    # TODO: If you prefer to use the CustomMachineType helper class, uncomment this code and comment the 2 lines above
    # Using CustomMachineType helper
    # cmt = CustomMachineType.from_str(instance.machine_type)
    # cmt.memory_mb = new_memory
    # cmt.extra_memory_used = True
    # instance.machine_type = str(cmt)
    operation = instance_client.update(
        project=project_id,
        zone=zone,
        instance=instance_name,
        instance_resource=instance,
    )
    wait_for_extended_operation(operation, "instance update")

    return instance_client.get(project=project_id, zone=zone, instance=instance_name)

What's next