1. Create a VM Compute Engine instance using the Datadog App and API Protection Service Extensions Callout container image.

   See Configuration for available environment variables when setting up your VM instance.

   Note: Be sure to update your Firewall rules to allow the Load Balancer and Datadog agent to communicate with the Callout VM instance.

2. Add the VM to an unmanaged instance group.

   Specify http:80 and grpc:443 (or your configured values) for the port mappings of the instance group.

3. Create a backend service with the following settings:

   • Protocol: HTTP2
   • Port name: grpc
   • Region: select your region
   • Health check port number: 80 (or your configured value)

4. Add the instance group with the service extension VM as a backend to this backend service.

5. Configure the Traffic Service Extension callout:

   1. In the Google Cloud console, go to Service Extensions and create a new Service Extension.
   2. Select your load balancer type.
   3. Select Traffic extensions as the type.
   4. Select your forwarding rules.
      
      

6. Create an Extension Chain

   1. To send all traffic to the extension, insert true in the Match condition.
   2. For Programability type, select Callouts.
   3. Select the backend service you created in the previous step.
   4. In Events, select only Request Headers and Response Headers. These are required. Do not select the bodies events. These are supported out-of-the-box by configuring the Service Extension callout with the body processing size limit.
   5. Optionally, enable the fail_open to still allow the traffic to pass through if the service extension fails or times out.
   
   
   Note: By default, if the service extension fails or times out, the proxy will return a 5xx error. To prevent this, enable the fail_open setting. When enabled, request or response processing continues without error even if the extension fails, ensuring your application remains available.

   Note: When body processing is enabled, the service extension decides based on its configuration whether to receive the request body, the response body, or both, along with the headers which are always included. This decision is made separately for requests and responses. Receiving body content can increase processing time, especially for large payloads. If you enable body processing, make sure to adjust your timeout settings to allow for the additional processing time required for body inspection and transfer.

이 구성을 완료하면 라이브러리가 애플리케이션에서 보안 데이터를 수집해 에이전트로 전송하고, 이 데이터는 다시 Datadog로 전송됩니다. 그러면 기본 감지 규칙에 기반해 공격자 기술과 잠재 구성 오류가 플래그되며, 이를 기반으로 문제 해결 단계를 진행할 수 있습니다.

3. 애플리케이션 보안 관리에서 감지 활동을 잘 하고 있는지 확인하려면 알려진 공격 패턴을 애플리케이션으로 보내세요. 예를 들어 다음 curl 스크립트가 포함된 파일을 실행해 보안 스캐너 감지됨 규칙을 트리거할 수 있습니다.

   for ((i=1;i<=250;i++)); 
   do
   # Target existing service’s routes
   curl https://your-application-url/existing-route -A dd-test-scanner-log;
   # Target non existing service’s routes
   curl https://your-application-url/non-existing-route -A dd-test-scanner-log;
   done

   참고: dd-test-scanner-log 값은 최신 릴리스에서 지원됩니다.

   애플리케이션을 활성화하고 실행한 몇 분 후 Application Signals Explorer에서 위협 정보가 표시되고 Vulnerability Explorer에 취약 정보가 표시됩니다.

You can use Terraform to automate the deployment of the App and API Protection GCP Service Extension. This simplifies the process of setting up the service extension to work with your existing load balancer.

Prerequisites for Terraform deployment

• Terraform installed on your local machine (version 1.0.0 or later)
• GCP credentials with appropriate permissions
• A Datadog API key (used to configure the Datadog Agent)
• An existing GCP Cloud Load Balancer for your application

Infrastructure Overview

The Terraform deployment will create the following components:

• A Datadog Agent VM for collecting traces with security events
• A VM running the Datadog Service Extension Callout in a container
• A firewall rule allowing communication between the extension and the Agent
• An unmanaged instance group containing the Service Extension VM
• A backend service configured for HTTP/2 with health checks
• A service extension connected to your existing load balancer

Deployment Steps

The App and API Protection Service Extension deployment requires several components that work together. We’ll create a Terraform module that encapsulates all these components, making the deployment process repeatable and easier to maintain.

1. Create a new directory and the necessary Terraform files:

   mkdir gcp-aap-service-extension && cd gcp-aap-service-extension
   touch main.tf variables.tf
   

2. Add the following code to your main.tf file. This file defines all the infrastructure components needed for the App and API Protection Service Extension, including network rules, VM instances, and load balancer configuration:

   # main.tf
   
   #----------------------------------------------------------
   # Network Configuration
   #----------------------------------------------------------
   
   # Firewall rule to allow the Service Extension to communicate with the Datadog Agent
   resource "google_compute_firewall" "aap_se_firewall" {
     name    = "${var.project_prefix}-dd-agent-firewall"
     network = "default"
   
     allow {
       protocol = "tcp"
       ports    = ["8126"]
     }
   
     source_tags = ["http-server"]
     target_tags = ["datadog-agent"]
   }
   
   #----------------------------------------------------------
   # Datadog Agent Configuration
   #----------------------------------------------------------
   
   # Datadog Agent container configuration
   module "gce-container-datadog-agent" {
     source = "terraform-google-modules/container-vm/google"
   
     container = {
       image = "public.ecr.aws/datadog/agent:latest"
       env = [
         {
           name = "DD_API_KEY",
           value = var.datadog_agent_api_key,
         },
         {
           name = "DD_ENV",
           value = "dev",
         },
       ]
     }
   }
   
   # Datadog Agent VM instance that collects traces from the Service Extension
   resource "google_compute_instance" "datadog_agent" {
     name         = "${var.project_prefix}-datadog-agent"
     machine_type = "e2-medium"
     zone         = var.zone
   
     boot_disk {
       auto_delete = true
   
       initialize_params {
         image = module.gce-container-datadog-agent.source_image
       }
   
     }
   
     network_interface {
       network    = "default"
       subnetwork = var.application_vpc_subnetwork
     }
   
     metadata = {
       gce-container-declaration = module.gce-container-datadog-agent.metadata_value
       google-logging-enabled    = "true"
     }
   
     lifecycle {
       create_before_destroy = true
     }
   
     tags = ["datadog-agent"]
   }
   
   #----------------------------------------------------------
   # Service Extension Callout Container Configuration
   #----------------------------------------------------------
   
   # Datadog App and API Protection GCP Service Extension container configuration
   module "gce-container-aap-service-extension" {
     source = "terraform-google-modules/container-vm/google"
   
     container = {
       image = "ghcr.io/datadog/dd-trace-go/service-extensions-callout:v2.2.2" # Replace with the latest version
       env = [
         {
           name = "DD_AGENT_HOST",
           value = google_compute_instance.datadog_agent.network_interface.0.network_ip,
         },
         # Enable the body processing by uncommenting the following lines. Set accordingly the body processing size limit.
         # {
         #   name = "DD_APPSEC_BODY_PARSING_SIZE_LIMIT",
         #   value = "10000000" # Body processing size limit in bytes. (10MB)
         # }
       ]
     }
   }
   
   # Service Extension VM instance (callout instance)
   resource "google_compute_instance" "default" {
     name         = "${var.project_prefix}-instance"
     machine_type = "e2-medium"
     zone         = var.zone
   
     boot_disk {
       auto_delete = true
   
       initialize_params {
         image = module.gce-container-aap-service-extension.source_image
       }
   
     }
   
     network_interface {
       network    = var.application_vpc_network
       subnetwork = var.application_vpc_subnetwork
     }
   
     metadata = {
       gce-container-declaration = module.gce-container-aap-service-extension.metadata_value
       google-logging-enabled    = "true"
     }
   
     lifecycle {
       create_before_destroy = true
     }
   
     # http-server: Allow access on the http server for health checks
     # https-server: Allow access on the 443 port for the AAP Service Extension
     tags = ["http-server", "https-server", "lb-health-check"]
   }
   
   #----------------------------------------------------------
   # Load Balancer Integration
   #----------------------------------------------------------
   
   # Unmanaged Instance Group including the App and API Protection Service Extension instance
   resource "google_compute_instance_group" "aap_se_instance_group" {
     name        = "${var.project_prefix}-instance-group"
     description = "Unmanaged instance group for the App and API Protection Service Extension"
     zone        = var.zone
   
     named_port {
       name = "http"
       port = 80
     }
   
     named_port {
       name = "grpc"
       port = "443"
     }
   
     instances = [
       google_compute_instance.default.self_link
     ]
   }
   
   # Health Check for the Backend Service
   resource "google_compute_health_check" "aap_se_health_check" {
     name                = "${var.project_prefix}-health-check"
     check_interval_sec  = 5
     timeout_sec         = 5
     healthy_threshold   = 2
     unhealthy_threshold = 2
   
     http_health_check {
       port         = 80
       request_path = "/"
     }
   }
   
   # Backend Service that points to the Service Extension instance group
   resource "google_compute_backend_service" "se_backend_service" {
     name                  = "${var.project_prefix}-backend-service"
     port_name             = "grpc"
     protocol              = "HTTP2"
     timeout_sec           = 10
     health_checks         = [google_compute_health_check.aap_se_health_check.self_link]
     load_balancing_scheme = "EXTERNAL_MANAGED"
   
     backend {
       group = google_compute_instance_group.aap_se_instance_group.self_link
     }
   }
   
   #----------------------------------------------------------
   # GCP Service Extension
   #----------------------------------------------------------
   
   # GCP Service Extension configuration for traffic interception
   resource "google_network_services_lb_traffic_extension" "default" {
     name        = "${var.project_prefix}-service-extension"
     description = "Datadog App and API Protection Service Extension"
     location    = "global"
   
     load_balancing_scheme = "EXTERNAL_MANAGED"
     forwarding_rules      = [var.load_balancer_forwarding_rule]
   
     extension_chains {
       name = "${var.project_prefix}-service-extension-chain"
   
       match_condition {
         cel_expression = "true" # Match all traffic
       }
   
       extensions {
         name      = "${var.project_prefix}-service-extension-chain-ext"
         service   = google_compute_backend_service.se_backend_service.self_link
   
         # Required: Please set your application hostname.
         authority = "datadoghq.com"
   
         # By default, if the service extension fails or times out, the proxy will return a 500 error.
         # To prevent this, enable the fail_open setting.
         # When enabled, the request processing is stopped when an error occurs but the request is not dropped, ensuring the availability of the application.
         fail_open = true
   
         # Mandatory, do not touch: Only set the Request and Response Headers events.
         # If bodies events are selected, the service extension will transfer and analyze the bodies for every request without applying relevant selection rules.
         # Please use the DD_APPSEC_BODY_PARSING_SIZE_LIMIT environment variable on the callout container to enable body processing.
         supported_events = ["REQUEST_HEADERS", "RESPONSE_HEADERS"]
   
         # Adjust your timeout settings depending of your processing needs (e.g. with body processing)
         # Note: This is the same option as the "message_timeout" for an Envoy ext_proc configuration.
         timeout   = "0.5s"
       }
     }
   }
   

3. Add the following content to the variables.tf file. This file defines all the required input variables for your Terraform configuration:

   # variables.tf
   
   variable "region" {
     description = "The GCP region where resources will be created (e.g., us-central1)"
     type        = string
     validation {
       condition     = length(var.region) > 0
       error_message = "Region cannot be empty."
     }
   }
   
   variable "zone" {
     description = "The GCP zone where zonal resources will be created (e.g., us-central1-a)"
     type        = string
     validation {
       condition     = length(var.zone) > 0
       error_message = "Zone cannot be empty."
     }
   }
   
   # Project configuration
   variable "project_prefix" {
     description = "Prefix for the project. All resource names will be prefixed with this value"
     type        = string
     validation {
       condition     = length(var.project_prefix) > 0
       error_message = "Project prefix cannot be empty."
     }
   }
   
   # Network configuration
   variable "application_vpc_network" {
   
     description = "Name of the VPC network for the application"
     type        = string
     validation {
       condition     = length(var.application_vpc_network) > 0
       error_message = "VPC network name cannot be empty."
     }
   }
   
   variable "application_vpc_subnetwork" {
   
     description = "Name of the VPC subnetwork for the application"
     type        = string
     validation {
       condition     = length(var.application_vpc_subnetwork) > 0
       error_message = "VPC subnetwork name cannot be empty."
     }
   }
   
   # Authentication and API keys
   variable "datadog_agent_api_key" {
     description = "Datadog API key"
     type        = string
     sensitive   = true
     validation {
       condition     = length(var.datadog_agent_api_key) > 0
       error_message = "Datadog API key cannot be empty."
     }
   }
   
   # Load balancer configuration
   variable "load_balancer_forwarding_rule" {
     description = "Self link to the forwarding rule for the load balancer"
   }
   

4. Include the module in your main Terraform project. This example shows how to reference the module you created above:

   # main.tf
   
   module "service_extension" {
     source                        = "./gcp-aap-service-extension"
     zone                          = "us-central1-a"
     region                        = "us-central1"
     project_prefix                = "datadog-aap"
     application_vpc_subnetwork    = "your-subnet-name"
     datadog_agent_api_key         = "your-datadog-api-key"
     load_balancer_forwarding_rule = "projects/your-project/regions/us-central1/forwardingRules/your-lb-rule" # or with a self link on your resource
   }
   

5. Deploy the infrastructure by running these commands in the directory where your Terraform files are located:

   terraform init
   terraform plan
   terraform apply
   

Post-deployment validation

The service extension automatically inspects all traffic passing through your load balancer for security threats.

이 구성을 완료하면 라이브러리가 애플리케이션에서 보안 데이터를 수집해 에이전트로 전송하고, 이 데이터는 다시 Datadog로 전송됩니다. 그러면 기본 감지 규칙에 기반해 공격자 기술과 잠재 구성 오류가 플래그되며, 이를 기반으로 문제 해결 단계를 진행할 수 있습니다.

3. 애플리케이션 보안 관리에서 감지 활동을 잘 하고 있는지 확인하려면 알려진 공격 패턴을 애플리케이션으로 보내세요. 예를 들어 다음 curl 스크립트가 포함된 파일을 실행해 보안 스캐너 감지됨 규칙을 트리거할 수 있습니다.

   for ((i=1;i<=250;i++)); 
   do
   # Target existing service’s routes
   curl https://your-application-url/existing-route -A dd-test-scanner-log;
   # Target non existing service’s routes
   curl https://your-application-url/non-existing-route -A dd-test-scanner-log;
   done

   참고: dd-test-scanner-log 값은 최신 릴리스에서 지원됩니다.

   애플리케이션을 활성화하고 실행한 몇 분 후 Application Signals Explorer에서 위협 정보가 표시되고 Vulnerability Explorer에 취약 정보가 표시됩니다.