このコマンドは、内部 IP アドレスとパブリック IP アドレスの両方を持つ VM を作成します（--no-address は VM 作成サンプル コマンドで指定されていません）。これにより、VM は Container Registry にアクセスしてプロキシ イメージを取得できます。内部 IP を使用できますが、VM が配置されているサブネットの限定公開の Google アクセスの有効化も行う必要があります。
次の環境変数をエクスポートします。
exportPROJECT=CUSTOMER_PROJECTexportREGION=VM_REGIONexportZONE=`gcloudcomputezoneslist--filter="name=${REGION}"--limit1--uri--project=${PROJECT}|sed's/.*\///'`exportNETWORK=CUSTOMER_VPC_NETWORK_NAMEexportSUBNET=CUSTOMER_VPC_SUBNET_NAMEexportINSTANCE_NAME=COMPUTE_ENGINE_VM_NAMEexportSQL_CONN=SQL_INSTANCE_CONNECTION_NAMEexportCDF_IP_RANGE=CLOUD_DATA_FUSION_IP_RANGEexportVM_IMAGE=$(gcloudcomputeimageslist--project=$PROJECT--filter=family:cos-stable--format='value(selfLink.scope())'--limit=1)exportSQL_PORT=DB_PORT# MySQL 3306 # PostgreSQL 5432 # SQLServer 1433
以下を置き換えます。
CUSTOMER_PROJECT: プロジェクトの名前
VM_REGION: Compute Engine VM が配置されているリージョン
CUSTOMER_VPC_NETWORK_NAME: ネットワークの名前
CUSTOMER_VPC_SUBNET_NAME: サブネットワークの名前
COMPUTE_ENGINE_VM_NAME: Compute Engine VM の名前
SQL_INSTANCE_CONNECTION_NAME: 前のステップで取得した接続名
CLOUD_DATA_FUSION_IP_RANGE: IP アドレス範囲
DB_PORT: Cloud SQL データベースのポート番号
次の gcloud CLI コマンドを使用して、Cloud Data Fusion の上り（内向き）トラフィックを許可するファイアウォール ルールを作成します。
[[["わかりやすい","easyToUnderstand","thumb-up"],["問題の解決に役立った","solvedMyProblem","thumb-up"],["その他","otherUp","thumb-up"]],[["わかりにくい","hardToUnderstand","thumb-down"],["情報またはサンプルコードが不正確","incorrectInformationOrSampleCode","thumb-down"],["必要な情報 / サンプルがない","missingTheInformationSamplesINeed","thumb-down"],["翻訳に関する問題","translationIssue","thumb-down"],["その他","otherDown","thumb-down"]],["最終更新日 2025-09-04 UTC。"],[[["\u003cp\u003eThis guide explains how to securely connect a private Cloud Data Fusion instance to a private Cloud SQL instance, such as MySQL, using a proxy Compute Engine VM.\u003c/p\u003e\n"],["\u003cp\u003eConnecting private instances of Cloud Data Fusion and Cloud SQL is highly recommended for enhanced security, and requires the use of VPC peering between the Cloud Data Fusion network and your project's VPC.\u003c/p\u003e\n"],["\u003cp\u003eA proxy Compute Engine VM is essential because the Cloud SQL network and Cloud Data Fusion network are not directly peered and cannot communicate transitively.\u003c/p\u003e\n"],["\u003cp\u003eThe process involves creating private instances in both Cloud Data Fusion and Cloud SQL, setting up a proxy VM with a static internal IP, and configuring firewall rules to allow traffic from Cloud Data Fusion.\u003c/p\u003e\n"],["\u003cp\u003eWhile the document emphasizes private instance connections, it also briefly outlines other connection possibilities, such as using public instances, although these options are less secure.\u003c/p\u003e\n"]]],[],null,["# Connect to a Cloud SQL-MySQL source\n\nThis page describes how to connect a private Cloud Data Fusion instance to a\nprivate Cloud SQL instance that hosts a MySQL database. A similar setup\nworks for Postgres and SQL Server instances. [Public connections are possible](#other-connections).\n| **Note:** [VPC peering](/data-fusion/docs/how-to/create-private-ip#set-up-vpc-peering) is used to connect the Cloud Data Fusion network to your project's VPC. You can connect the Cloud SQL network to your project's VPC using VPC peering or VPN. \n| This page assumes that you're using VPC peering to connect the Cloud SQL network to your project's VPC. If you're using VPN, then don't perform the tasks described on this page. To establish the connection, you can use the internal IP address of the Cloud SQL instance in Cloud Data Fusion.\n\n**Recommended**: For security reasons, use\nprivate instances of Cloud SQL and Cloud Data Fusion.\n\nCreate private instances in Cloud Data Fusion and Cloud SQL\n-----------------------------------------------------------\n\nTo connect to a private Cloud SQL instance from a private\nCloud Data Fusion instance, you use a proxy Compute Engine VM. A proxy\nis required because the Cloud SQL network is not directly peered with\nthe Cloud Data Fusion network, and transitive peers cannot communicate with\neach other (see the [VPC Network Peering overview](/vpc/docs/vpc-peering)).\n\nTo create the private instances, follow these steps:\n\n1. If you don't already have one, [create a private Cloud Data Fusion\n instance](/data-fusion/docs/how-to/create-private-ip). This includes:\n\n 1. Setting up a VPC network\n 2. Allocating an IP range\n2. [Create a private Cloud SQL instance](/sql/docs/mysql/create-instance)\n in the same VPC network as your Cloud Data Fusion instance.\n\n3. After the instance is created, go to the Cloud SQL **Instances**\n page and expand the **Instance details** . In the **Connect to this\n instance** section, copy the IP and the connection name.\n\n [Go to Cloud SQL Instances](https://console.cloud.google.com/sql/instances)\n\nCreate a private Compute Engine VM\n----------------------------------\n\nTo create the VM, you can enter the following commands in\n[Cloud Shell](/shell) or another environment where the\n[Google Cloud SDK](/sdk/docs) is installed.\n\nThe commands create a VM with both internal and public IP addresses\n(`--no-address` isn't specified in the sample VM creation command) so that the\nVM can access the Container Registry to get the proxy image. You can use internal\nIPs, but you also must\n[enable Private Google Access](/vpc/docs/configure-private-google-access) for\nthe subnet where the VM is located.\n\n1. Export the following environmental variables:\n\n export PROJECT=\u003cvar translate=\"no\"\u003eCUSTOMER_PROJECT\u003c/var\u003e\n export REGION=\u003cvar translate=\"no\"\u003eVM_REGION\u003c/var\u003e\n export ZONE=`gcloud compute zones list --filter=\"name=${REGION}\" --limit 1 --uri --project=${PROJECT}| sed 's/.*\\///'`\n export NETWORK=\u003cvar translate=\"no\"\u003eCUSTOMER_VPC_NETWORK_NAME\u003c/var\u003e\n export SUBNET=\u003cvar translate=\"no\"\u003eCUSTOMER_VPC_SUBNET_NAME\u003c/var\u003e\n export INSTANCE_NAME=\u003cvar translate=\"no\"\u003eCOMPUTE_ENGINE_VM_NAME\u003c/var\u003e\n export SQL_CONN=\u003cvar translate=\"no\"\u003eSQL_INSTANCE_CONNECTION_NAME\u003c/var\u003e\n export CDF_IP_RANGE=\u003cvar translate=\"no\"\u003eCLOUD_DATA_FUSION_IP_RANGE\u003c/var\u003e\n export VM_IMAGE=$(gcloud compute images list --project=$PROJECT --filter=family:cos-stable --format='value(selfLink.scope())' --limit=1)\n export SQL_PORT=\u003cvar translate=\"no\"\u003eDB_PORT\u003c/var\u003e # MySQL 3306 # PostgreSQL 5432 # SQLServer 1433\n\n Replace the following:\n - \u003cvar translate=\"no\"\u003eCUSTOMER_PROJECT\u003c/var\u003e: Your project's name\n - \u003cvar translate=\"no\"\u003eVM_REGION\u003c/var\u003e: The region where the Compute Engine VM is located\n - \u003cvar translate=\"no\"\u003eCUSTOMER_VPC_NETWORK_NAME\u003c/var\u003e: Your network's name\n - \u003cvar translate=\"no\"\u003eCUSTOMER_VPC_SUBNET_NAME\u003c/var\u003e: Your subnetwork's name\n - \u003cvar translate=\"no\"\u003eCOMPUTE_ENGINE_VM_NAME\u003c/var\u003e: The name of the Compute Engine VM\n - \u003cvar translate=\"no\"\u003eSQL_INSTANCE_CONNECTION_NAME\u003c/var\u003e: The connection name from the previous step\n - \u003cvar translate=\"no\"\u003eCLOUD_DATA_FUSION_IP_RANGE\u003c/var\u003e: The IP address range\n - \u003cvar translate=\"no\"\u003eDB_PORT\u003c/var\u003e: The port number of the Cloud SQL database\n2. Create a firewall rule to allow Cloud Data Fusion ingress traffic with\n the following gcloud CLI command:\n\n gcloud compute firewall-rules create allow-private-cdf \\\n --allow=tcp:22,tcp:${SQL_PORT} \\\n --source-ranges=$CDF_IP_RANGE --network=$NETWORK --project=$PROJECT\n\n3. Create the VM with the following gcloud CLI command:\n\n gcloud compute --project=${PROJECT} instances create ${INSTANCE_NAME} \\\n --zone=${ZONE} \\\n --machine-type=g1-small \\\n --subnet=${SUBNET} \\\n --metadata=startup-script=\"docker run -d -p 0.0.0.0:${SQL_PORT}:${SQL_PORT} gcr.io/cloudsql-docker/gce-proxy:latest /cloud_sql_proxy -instances=${SQL_CONN}=tcp:0.0.0.0:${SQL_PORT}\" \\\n --maintenance-policy=MIGRATE \\\n --scopes=https://www.googleapis.com/auth/cloud-platform \\\n --image=${VM_IMAGE} \\\n --image-project=cos-cloud\n\n4. Get the VM internal IP with the following gcloud CLI command:\n\n export IP=`gcloud compute \\\n --project=${PROJECT} instances describe ${INSTANCE_NAME} \\\n --zone ${ZONE} | grep \"networkIP\" | awk '{print $2}'`\n\n5. Promote the VM internal IP to a static IP with the following\n gcloud CLI command:\n\n gcloud compute --project=${PROJECT} addresses create mysql-proxy \\\n --addresses ${IP} --region ${REGION} --subnet ${SUBNET}\n\n6. Get the IP to be used in Cloud Data Fusion MySQL JDBC connection string\n when you set up the connection:\n\n echo ${IP}\n\nAfter the VM is created with a static IP, use the IP as the host or enter it in\nthe JDBC connection string to access the MySQL database from\nCloud Data Fusion.\n\nAccess MySQL from within Cloud Data Fusion\n------------------------------------------\n\nBefore you can connect to the MySQL instance from the Cloud Data Fusion\ninstance, install the MySQL JDBC driver from the Cloud Data Fusion Hub (or\ndirectly from\n[MySQL community downloads](https://dev.mysql.com/downloads/file/?id=462850)).\n| **Note:** In this case, use the general MySQL driver instead of Cloud SQL-MySQL driver.\n\nEnter the connection details for your MySQL database in\nCloud Data Fusion. You can enter the details and test the connection on\nthe **Wrangler** page.\n\n### Optional: Other ways to connect to MySQL\n\nYou can access MySQL in the following ways in Cloud Data Fusion:\n\n- On the **Studio** page, click the MySQL source plugin and enter the connection details.\n- On the **Studio** page, instead use the Database source plugin, which has\n similar properties (**Connection string** , **Username** , **Password** , and\n **Connection Arguments**).\n\n | **Note:** Unlike the MySQL plugins, the Database plugin isn't designed specifically for MySQL connections. You must understand the JDBC string format to add the connection details.\n\nOther connections\n-----------------\n\nAlthough connecting private instances is recommended, the following connections\nare possible:\n\n- A public Cloud Data Fusion instance to a public Cloud SQL instance (by using a public Cloud SQL proxy or allowlisting the `0.0.0.0/0` range in an authorized network). See the [public IP connection option](https://cloud.google.com/sql/docs/mysql/connect-overview#public_ip) for Cloud SQL.\n- A private Cloud Data Fusion instance and public Cloud SQL instance (by using a public Cloud SQL proxy in an authorized network).\n- A public Cloud Data Fusion instance and a private Cloud SQL\n instance."]]
