Contact UsGet started

Running an instance group with autoscaling using the management console, CLI, and API

Written by
Updated at November 14, 2024

To configure an instance group with an autoscaling policy:

  1. Prepare your cloud.
  2. Prepare the environment.
  3. Create an instance group with autoscaling and a network load balancer.
  4. Add a network load balancer with a target group.
  5. Test your instance group and network load balancer.

If you no longer need the resources you created, delete them.

Prepare your cloud

Sign up for Yandex Cloud and create a billing account:

  1. Go to the management console and log in to Yandex Cloud or create an account if you do not have one yet.
  2. On the Yandex Cloud Billing page, make sure you have a billing account linked and it has the ACTIVE or TRIAL_ACTIVE status. If you do not have a billing account, create one.

If you have an active billing account, you can go to the cloud page to create or select a folder for your infrastructure to operate in.

Learn more about clouds and folders.

  1. If you don't have the Yandex Cloud command line interface yet, install it.

  2. For the scripts from the tutorial to run correctly, download and install the jq utility.

  3. To check auto scaling, install the wrk benchmarking tool.

The infrastructure cost includes:

Prepare the environment

  1. Create a service account named for-autoscale and assign it the editor role:

    1. In the management console, select the folder where you want to create your service account.
    2. In the list of services, select Identity and Access Management.
    3. Click Create service account. In the window that opens:
      • In the Name field, specify for-autoscale.
      • To assign the service account the editor role for the current folder, click Add role and select editor.
      • Click Create.

    1. Create a service account:

      The folder specified in the CLI profile is used by default. You can specify a different folder using the --folder-name or --folder-id parameter.

      yc iam service-account create --name for-autoscale

      Result:

      id: ajelabcde12f********
folder_id: b0g12ga82bcv********
created_at: "2020-11-30T14:32:18.900092Z"
name: for-autoscale

    2. Assign the role to the service account:

      yc resource-manager folder add-access-binding b1g23ga82bcv******** \
  --role editor \
  --subject serviceAccount:ajelabcde12f********

    Use the create REST API method for the ServiceAccount resource or the ServiceAccountService/Create gRPC API call.

  2. Create a network named yc-auto-network and subnets in two availability zones:

    1. In the management console, select the folder where you want to create a network.
    2. Select Virtual Private Cloud.
    3. Click Create network.
    4. In the Name field, enter yc-auto-network as the network name.
    5. In the Advanced field, enable the Create subnets option.
    6. Click Create network.

    1. Create a network:

      yc vpc network create --name yc-auto-network

      Result:

      id: enpabce123hd********
folder_id: b0g12ga82bcv********
created_at: "2020-11-30T14:57:15Z"
name: yc-auto-network

    2. Create a subnet in the ru-central1-a availability zone:

      yc vpc subnet create \
  --network-id enpabce123hd******** \
  --range 192.168.1.0/24 \
  --zone ru-central1-a

      Result:

      id: e1lnabc23r1c********
folder_id: b0g12ga82bcv********
created_at: "2020-11-30T16:23:12Z"
network_id: enpabce123hd********
zone_id: ru-central1-a
v4_cidr_blocks:
- 192.168.1.0/24

    3. Create a subnet in the ru-central1-b availability zone:

      yc vpc subnet create \
  --network-id enpabce123hd******** \
  --range 192.168.2.0/24 \
  --zone ru-central1-b

      Result:

      id: b1csa2b3clid********
folder_id: b0g12ga82bcv********
created_at: "2020-11-30T16:25:02Z"
network_id: enpabce123hd********
zone_id: ru-central1-b
v4_cidr_blocks:
- 192.168.2.0/24

    1. Create a network.

      Use the create REST API method for the Network resource or the NetworkService/Create gRPC API call.

    2. Create subnets in the ru-central1-a and ru-central1-b availability zones.

      Use the create REST API method for the Subnet resource or the SubnetService/Create gRPC API call.

  3. Create a security group:

    1. In the management console, select Virtual Private Cloud.
    2. Open the Security groups tab.
    3. Create a security group for the load balancer:

      1. Click Create group.

      2. Enter a Name for the group: sg-autoscale.

      3. Select yc-auto-network for Network.

      4. Under Rules, create rules using the instructions below the table:

        Traffic
        direction        		 Description Port
        range        		 Protocol Source /
        target type        		 Source /
        target
        Outgoing any All Any CIDR 0.0.0.0/0
        Incoming ext-http 80 TCP CIDR 0.0.0.0/0
        Incoming healthchecks 80 TCP Load balancer health checks —
        1. Select the Outgoing traffic or Incoming traffic tab.
        2. Click Add rule.
        3. In the Port range field of the window that opens, specify a single port or a range of ports that traffic will come to or from.
        4. In the Protocol field, specify the appropriate protocol or leave Any to allow traffic transmission over any protocol.
        5. In the Purpose or Source field, select the purpose of the rule:
          • CIDR: Rule will apply to the range of IP addresses. In the CIDR blocks field, specify the CIDR and masks of subnets that traffic will come to or from. To add multiple CIDRs, click Add CIDR.
          • Security group: Rule will apply to the VMs from the current group or the selected security group.
          • Load balancer health checks: Rule allowing a load balancer to health check VMs.
        6. Click Save. Repeat the steps to create all the rules from the table.

      5. Click Save.

Create an instance group with autoscaling and a network load balancer

  1. All instance groups are created from Container Optimized Image. Each instance contains a Docker container running a web server that emulates the service utilization.

    Find out the ID of the latest version of the public Container Optimized Image.

    A Container Optimized Image in a Container Registry registry may be updated and changed with new releases. This will not automatically update the VM image to the latest version. To create an instance group with the latest Container Optimized Image version, you need to check whether it is available yourself:

    yc compute image get-latest-from-family container-optimized-image --folder-id standard-images

    Result:

    id: <ID_of_latest_COI_version>
folder_id: standard-images
...
    1. Go to the Cloud Marketplace page and select the image with the configuration you need:
    2. Under Product IDs, copy the value of image_id.

  2. Save the specification of the instance group with network load balancer to the specification.yaml file:

    name: auto-group
service_account_id: <service_account_ID>
scale_policy:
  auto_scale:
    min_zone_size: 1
    max_size: 6
    measurement_duration: 60s
    warmup_duration: 120s
    stabilization_duration: 60s
    initial_size: 2
    cpu_utilization_rule:
      utilization_target: 40
deploy_policy:
  max_unavailable: 1
allocation_policy:
  zones:
    - zone_id: ru-central1-a
    - zone_id: ru-central1-b
load_balancer_spec:
  target_group_spec:
    name: auto-group-tg
instance_template:
  service_account_id: <service_account_ID>
  platform_id: standard-v3
  resources_spec:
    memory: 2G
    cores: 2
    core_fraction: 100
  metadata:
    docker-container-declaration: |-
      spec:
        containers:
        - image: cr.yandex/yc/demo/web-app:v1
          securityContext:
            privileged: false
          tty: false
          stdin: false
  boot_disk_spec:
    mode: READ_WRITE
    disk_spec:
      type_id: network-hdd
      size: 30G
      image_id: fd8iv792kira******** # ID of the public . Container Optimized Image
  network_interface_specs:
    - network_id: <cloud_network_ID>
      primary_v4_address_spec: { one_to_one_nat_spec: { ip_version: IPV4 }}

  3. In specification.yaml, replace the values in angle brackets for the real values you got in the previous steps.

  4. Create an instance group named auto-group using the specification.yaml specification:

    Run this command:

    yc compute instance-group create --file=specification.yaml

    Result:

    done (2m45s)
id: cl0hmabc1nd2********
folder_id: b0g12ga82bcv********
...
  target_group_id: enpoi5jhfps3********
...
    name: auto-group-tg
service_account_id: ajelabcde12f********
status: ACTIVE

    Use the CreateFromYaml method for the InstanceGroup resource.

  5. Make sure that the instance group was created:

    1. In the management console, select the folder where you created the instance group.
    2. Select Compute Cloud.
    3. In the left-hand panel, click Instance groups.
    4. Select the auto-group instance group.
    yc compute instance-group list-instances auto-group

    Result:

    +----------------------+---------------------------+----------------+--------------+------------------------+----------------+
|     INSTANCE ID      |           NAME            |  EXTERNAL IP   | INTERNAL IP  |         STATUS         | STATUS MESSAGE |
+----------------------+---------------------------+----------------+--------------+------------------------+----------------+
| epdab0c1ji2a******** | cl0habce1nd2********-fned | 84.201.163.202 | 192.168.1.34 | RUNNING_ACTUAL [4m26s] |                |
| ef0uabc1s2fb******** | cl0habce1nd2********-craq | 130.193.56.102 | 192.168.2.19 | RUNNING_ACTUAL [4m14s] |                |
+----------------------+---------------------------+----------------+--------------+------------------------+----------------+

    To view the list of created instance groups, use the list REST API method for the InstanceGroup resource or the InstanceGroupService/List gRPC API call.

Add a network load balancer with a target group

  1. Create a network load balancer named group-balancer and connect it to the instance group you created:

    1. In the management console, select the folder you want to create a load balancer in.
    2. Select Network Load Balancer.
    3. Click Create a network load balancer.
    4. Enter group-balancer for name.
    5. In the Public address field, specify Auto.
    6. Under Listeners, click Add listener. In the window that opens, specify:
      • Name: http.
      • Port (port the load balancer will accept incoming traffic at): 80.
      • Target port (port the load balancer will route traffic to): 80.
      • Click Add.
    7. Under Target groups, click Add target group.
    8. In the Target group field, select the auto-group-tg instance group and click Configure. In the window that opens, specify:
      • Name: tcp.
      • Type: TCP.
      • Port: 80.
      • Click Apply.
    9. Click Create.
    yc load-balancer network-load-balancer create \
  --listener external-ip-version=ipv4,name=http,port=80,protocol=tcp,target-port=80 \
  --name group-balancer \
  --target-group healthcheck-name=tcp,healthcheck-tcp-port=80,target-group-id=enpoi5jhfps3********

    Result:

    done (16s)
id: b0rbabc1m2ed********
folder_id: b0g12ga82bcv********
...
    healthy_threshold: "2"
    tcp_options:
      port: "80"
    1. Create a load balancer using the create REST API method for the NetworkLoadBalancer resource or the NetworkLoadBalancerService/Create gRPC API call.
    2. Add a listener to the load balancer using the addListener REST API method for the NetworkLoadBalancer resource or the NetworkLoadBalancerService/AddListener gRPC API call.
    3. Connect the target group to the load balancer using the attachTargetGroup REST API method for the NetworkLoadBalancer resource or the NetworkLoadBalancerService/AttachTargetGroup gRPC API call.
    4. Add the load balancer to the instance group using the addTargets REST API method for the TargetGroup resource or the TargetGroupService/AddTargets gRPC API call.

  2. Check that the network load balancer named group-balancer has been created and linked to the instance group:

    1. In the management console, select the folder you created the load balancer in.
    2. Select Network Load Balancer.
    3. Select group-balancer.
    yc load-balancer network-load-balancer list

    Result:

    +----------------------+----------------+-----------------+----------+----------------+------------------------+--------+
|          ID          |      NAME      |    REGION ID    |   TYPE   | LISTENER COUNT | ATTACHED TARGET GROUPS | STATUS |
+----------------------+----------------+-----------------+----------+----------------+------------------------+--------+
| b0rbabc1m2ed******** | group-balancer | ru-central1 | EXTERNAL |              1 | b0rdabckribe********   | ACTIVE |
+----------------------+----------------+-----------------+----------+----------------+------------------------+--------+

    Use the list REST API method for the NetworkLoadBalancer resource or the NetworkLoadBalancerService/List gRPC API call.

Test your instance group and network load balancer

  1. Create a load on an instance.

    To do this, save the script named request.sh to the home directory. The script will send a request to one of the VMs through the group-balancer load balancer. The request will utilize 100% CPU for 30 seconds.

    EXTERNAL_IP=$(yc load-balancer network-load-balancer get group-balancer --format=json | jq -r .listeners[0].address)

curl "http://$EXTERNAL_IP/burn-cpu?time=30000&load=100"

  2. Run the script:

    sh request.sh

    Result:

    projects/b0g12ga82bcv********/zones/ru-central1-b

  3. View the instance utilization:

    1. In the management console, select the folder where you created the instance group.

    2. Select Compute Cloud.

    3. In the left-hand panel, click Instance groups.

    4. Select auto-group.

    5. Go to the Monitoring tab.

      The load balancer sent the request to an instance in the group. In the availability zone this instance belongs to, the average CPU utilization is higher than in other zones (see the Average CPU utilization in zone chart).

Test autoscaling

To test autoscaling for your instance group, increase the CPU utilization of each instance. In the specification.yaml file, the scale_policy.auto_scale.cpu_utilization_rule.utilization_target parameter is set to 40: it means that the target utilization level is 40% CPU. If you exceed the target utilization, the number of VMs in the group will increase.

  1. Increase the utilization for the instance group.

    To do this, save the script named load.sh to the home directory. The script sends requests to the instance group through 12 threads for 10 minutes. Each VM instance utilizes 20% CPU on each core that processes the request. The instance group utilization is 240% CPU at any given time. To be sure that requests are evenly distributed across the instances in the group, the script executes multiple parallel requests utilizing 20% CPU rather than one request utilizing 240% CPU.

    EXTERNAL_IP=$(yc load-balancer network-load-balancer get group-balancer --format=json | jq -r .listeners[0].address)

wrk -H "Connection: close" -t12 -c12 -d10m "http://$EXTERNAL_IP/burn-cpu?time=5000&load=20"

  2. Run the script:

    sh load.sh

    Result:

    Running 10m test @ http://130.193.56.111/burn-cpu?time=5000&load=20
  12 threads and 12 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
...
Requests/sec: 15.04
Transfer/sec: 3.20KB

  3. View the utilization levels:

    1. In the management console, select the folder where you created the auto-group instance group.
    2. Select Compute Cloud.
    3. In the panel on the left, click Instance groups.
    4. Select auto-group.
    5. Go to the Monitoring tab.
      The chart Number of instances in zone shows how the number of instances changed in each availability zone. The chart Average CPU utilization in zone shows average CPU utilization in each availability zone.
    6. Go to the Logs tab.
      The page displays messages relating to the instance group autoscaling.

    The total utilization of 240% CPU was evenly distributed between two instances in two availability zones and exceeded the target utilization of 40% CPU. Yandex Compute Cloud created one instance more in each availability zone to result in four instances in the group. When the script stopped utilizing the CPU, Compute Cloud automatically decreased the number of instances in the group to two.

How to delete the resources you created

To stop paying for the resources you created:

  1. Delete the network load balancer.
  2. Delete the instance group.

See also

Previous
Overview
Next
Terraform