Tentang enkripsi sisi klien

MySQL | PostgreSQL | SQL Server

Halaman ini menjelaskan cara mengimplementasikan enkripsi sisi klien di Cloud SQL.

Ringkasan

Enkripsi sisi klien adalah tindakan mengenkripsi data sebelum menulisnya ke Cloud SQL. Anda dapat mengenkripsi data Cloud SQL dengan cara yang hanya dapat didekripsi oleh aplikasi Anda.

Untuk mengaktifkan enkripsi sisi klien, Anda memiliki opsi berikut ini:

Menggunakan kunci enkripsi yang disimpan di Cloud Key Management Service (Cloud KMS).
Menggunakan kunci enkripsi yang disimpan secara lokal di aplikasi Anda.

Dalam topik ini, kami menjelaskan cara menggunakan opsi pertama, yang memberikan opsi pengelolaan kunci paling lancar. Kami membuat kunci enkripsi di Cloud KMS dan menerapkan enkripsi menyeluruh menggunakan Tink, library kriptografi open source Google.

Mengapa Anda memerlukan enkripsi sisi klien?

Anda memerlukan enkripsi sisi klien jika ingin melindungi data Cloud SQL di tingkat kolom ¹. Bayangkan Anda memiliki tabel nama dan nomor kartu kredit. Anda ingin memberi pengguna akses ke tabel ini, tetapi tidak ingin mereka melihat nomor kartu kredit. Anda dapat mengenkripsi nomor menggunakan enkripsi sisi klien. Selama pengguna tidak diberi akses ke kunci enkripsi di Cloud KMS, mereka tidak dapat membaca informasi kartu kredit.

Membuat kunci menggunakan Cloud KMS

Cloud KMS memungkinkan Anda untuk membuat dan mengelola kunci di Google Cloud Platform.

Cloud KMS mendukung berbagai jenis kunci yang berbeda. Untuk enkripsi sisi klien, Anda perlu membuat kunci simetris.

Untuk memberi aplikasi Anda akses ke kunci di Cloud KMS, Anda perlu memberikan peran cloudkms.cryptoKeyEncrypterDecrypter ke akun layanan yang digunakan aplikasi Anda. Di gcloud, Anda menggunakan perintah berikut untuk melakukannya:

gcloud kms keys add-iam-policy-binding key \
--keyring=key-ring \
--location=location \
--member=serviceAccount:service-account-name@example.domain.com \
--role=roles/cloudkms.cryptoKeyEncrypterDecrypter

Meskipun Anda dapat menggunakan kunci KMS untuk langsung mengenkripsi data, di sini kami menggunakan solusi yang lebih fleksibel yang disebut enkripsi menyeluruh. Dengan begitu, kami dapat mengenkripsi pesan yang lebih panjang dari 64KB, yang merupakan ukuran pesan maksimum yang dapat didukung oleh Cloud Key Management Service API.

Enkripsi menyeluruh Cloud KMS

Dalam enkripsi menyeluruh, kunci KMS bertindak sebagai kunci enkripsi kunci (KEK). Artinya, hal ini digunakan untuk mengenkripsi kunci enkripsi data (DEK) yang kemudian digunakan untuk mengenkripsi data sebenarnya.

Setelah membuat KEK di Cloud KMS, untuk mengenkripsi setiap pesan, Anda perlu:

Membuat kunci enkripsi data (DEK) secara lokal.
Gunakan DEK ini secara lokal untuk mengenkripsi pesan.
Memanggil Cloud KMS untuk mengenkripsi (menggabungkan) DEK dengan KEK.
Menyimpan data terenkripsi dan DEK yang digabungkan.

Alih-alih menerapkan enkripsi menyeluruh dari awal, kami menggunakan Tink dalam topik ini.

Tink

Tink adalah library multi-bahasa dan lintas platform yang menyediakan API kriptografi tingkat tinggi. Untuk mengenkripsi data dengan enkripsi menyeluruh Tink, Anda perlu menyediakan Tink dengan URI kunci yang mengarah ke KEK Anda di Cloud KMS, dan kredensial yang memungkinkan Tink untuk menggunakan KEK. Tink menghasilkan DEK, mengenkripsi data, menggabungkan DEK, dan menampilkan satu ciphertext dengan data terenkripsi dan DEK yang digabungkan.

Tink mendukung enkripsi menyeluruh di C++, Java, Go, dan Python menggunakan AEAD API:

public interface Aead{
  byte[] encrypt(final byte[] plaintext, final byte[] associatedData)
  throws…
  byte[] decrypt(final byte[] ciphertext, final byte[] associatedData)
  throws…
}

Selain argumen pesan/ciphertext normal, metode enkripsi dan dekripsi mendukung data terkait yang bersifat opsional. Argumen ini dapat digunakan untuk mengaitkan ciphertext ke sebuah data. Misalnya, Anda memiliki database dengan kolom user-id dan kolom encrypted-medical-history. Dalam hal ini, kolom user-id mungkin harus digunakan sebagai data terkait saat mengenkripsi histori medis. Hal ini memastikan bahwa penyerang tidak dapat memindahkan histori medis dari satu pengguna ke pengguna lain. Hal ini juga digunakan untuk memverifikasi bahwa Anda memiliki baris data yang benar saat menjalankan kueri.

Sampel

Di bagian ini, kami akan membahas kode contoh untuk database informasi pemilih yang menggunakan enkripsi sisi klien. Kode contoh menunjukkan cara:

Membuat tabel database dan kumpulan koneksi
Menyiapkan Tink untuk enkripsi menyeluruh
Mengenkripsi dan mendekripsi data menggunakan enkripsi menyeluruh Tink dengan KEK di Cloud KMS

Sebelum memulai

Buatlah Instance Cloud SQL dengan mengikuti instructions ini. Catatlah string koneksi, pengguna database, dan sandi database yang Anda buat.
Buatlah database untuk aplikasi Anda dengan mengikuti instructions ini. Catatlah nama database.
Buatlah kunci KMS untuk aplikasi Anda dengan mengikuti instructions ini. Salinlah nama resource dari kunci yang Anda buat.
Buatlah akun layanan dengan izin 'Klien Cloud SQL' dengan mengikuti instructions ini.
Tambahkan izin 'Encrypter/Decrypter Cloud KMS CryptoKey' untuk kunci tersebut ke akun layanan Anda dengan mengikuti instructions ini.

Buatlah kumpulan koneksi dan buat tabel baru dalam database.

Java


import com.zaxxer.hikari.HikariConfig;
import com.zaxxer.hikari.HikariDataSource;
import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.SQLException;
import javax.sql.DataSource;

public class CloudSqlConnectionPool {

  public static DataSource createConnectionPool(String dbUser, String dbPass, String dbName,
      String instanceConnectionName) {
    HikariConfig config = new HikariConfig();
    config.setDataSourceClassName("com.microsoft.sqlserver.jdbc.SQLServerDataSource");
    config.setUsername(dbUser); // e.g. "root", "sqlserver"
    config.setPassword(dbPass); // e.g. "my-password"
    config.addDataSourceProperty("databaseName", dbName);

    // The Cloud SQL Java Connector provides SSL encryption so
    // it should be disabled at the driver level
    config.addDataSourceProperty("encrypt", "false");

    config.addDataSourceProperty("socketFactoryClass",
        "com.google.cloud.sql.sqlserver.SocketFactory");
    config.addDataSourceProperty("socketFactoryConstructorArg", instanceConnectionName);
    DataSource pool = new HikariDataSource(config);
    return pool;
  }

  public static void createTable(DataSource pool, String tableName) throws SQLException {
    // Safely attempt to create the table schema.
    try (Connection conn = pool.getConnection()) {

      String stmt = String.format("IF NOT EXISTS("
          + "SELECT * FROM sysobjects WHERE name='%s' and xtype='U')"
          + "CREATE TABLE %s ("
          + "vote_id INT NOT NULL IDENTITY,"
          + "time_cast DATETIME NOT NULL,"
          + "team VARCHAR(6) NOT NULL,"
          + "voter_email VARBINARY(255)"
          + "PRIMARY KEY (vote_id));", tableName, tableName);
      try (PreparedStatement createTableStatement = conn.prepareStatement(stmt);) {
        createTableStatement.execute();
      }
    }
  }
}

Python

import pytds
import sqlalchemy
from sqlalchemy import Column
from sqlalchemy import DateTime
from sqlalchemy import Integer
from sqlalchemy import Table

def init_tcp_connection_engine(
    db_user: str, db_pass: str, db_name: str, db_host: str
) -> sqlalchemy.engine.base.Engine:
    # Remember - storing secrets in plaintext is potentially unsafe. Consider using
    # something like https://cloud.google.com/secret-manager/docs/overview to help keep
    # secrets secret.

    # Extract host and port from db_host
    host_args = db_host.split(":")
    db_hostname, db_port = host_args[0], int(host_args[1])

    def connect_with_pytds() -> pytds.Connection:
        return pytds.connect(
            db_hostname,  # e.g. "127.0.0.1"
            user=db_user,  # e.g. "my-database-user"
            password=db_pass,  # e.g. "my-database-password"
            database=db_name,  # e.g. "my-database-name"
            port=db_port,  # e.g. 1433
            bytes_to_unicode=False,  # disables automatic decoding of bytes
        )

    pool = sqlalchemy.create_engine(
        # This allows us to use the pytds sqlalchemy dialect, but also set the
        # bytes_to_unicode flag to False, which is not supported by the dialect
        "mssql+pytds://",
        creator=connect_with_pytds,
    )

    print("Created TCP connection pool")
    return pool

def init_db(
    db_user: str,
    db_pass: str,
    db_name: str,
    db_host: str,
    table_name: str,
) -> sqlalchemy.engine.base.Engine:
    db = init_tcp_connection_engine(db_user, db_pass, db_name, db_host)

    # Create tables (if they don't already exist)
    if not db.has_table(table_name):
        metadata = sqlalchemy.MetaData(db)
        Table(
            table_name,
            metadata,
            Column("vote_id", Integer, primary_key=True, nullable=False),
            Column("voter_email", sqlalchemy.types.VARBINARY, nullable=False),
            Column("time_cast", DateTime, nullable=False),
            Column("team", sqlalchemy.types.VARCHAR(6), nullable=False),
        )
        metadata.create_all()

    print(f"Created table {table_name} in db {db_name}")
    return db

Melakukan inisialisasi AEAD primitif secara menyeluruh dengan Tink.

Java


import com.google.crypto.tink.Aead;
import com.google.crypto.tink.KmsClient;
import com.google.crypto.tink.aead.AeadConfig;
import com.google.crypto.tink.aead.AeadKeyTemplates;
import com.google.crypto.tink.aead.KmsEnvelopeAead;
import com.google.crypto.tink.integration.gcpkms.GcpKmsClient;
import java.security.GeneralSecurityException;

public class CloudKmsEnvelopeAead {

  public static Aead get(String kmsUri) throws GeneralSecurityException {
    AeadConfig.register();

    // Create a new KMS Client
    KmsClient client = new GcpKmsClient().withDefaultCredentials();

    // Create an AEAD primitive using the Cloud KMS key
    Aead gcpAead = client.getAead(kmsUri);

    // Create an envelope AEAD primitive.
    // This key should only be used for client-side encryption to ensure authenticity and integrity
    // of data.
    return new KmsEnvelopeAead(AeadKeyTemplates.AES128_GCM, gcpAead);
  }
}

Python

import logging

import tink
from tink import aead
from tink.integration import gcpkms

logger = logging.getLogger(__name__)

def init_tink_env_aead(key_uri: str, credentials: str) -> tink.aead.KmsEnvelopeAead:
    aead.register()

    try:
        gcp_client = gcpkms.GcpKmsClient(key_uri, credentials)
        gcp_aead = gcp_client.get_aead(key_uri)
    except tink.TinkError as e:
        logger.error("Error initializing GCP client: %s", e)
        raise e

    # Create envelope AEAD primitive using AES256 GCM for encrypting the data
    # This key should only be used for client-side encryption to ensure authenticity and integrity
    # of data.
    key_template = aead.aead_key_templates.AES256_GCM
    env_aead = aead.KmsEnvelopeAead(key_template, gcp_aead)

    print(f"Created envelope AEAD Primitive using KMS URI: {key_uri}")

    return env_aead

Mengenkripsi data dan memasukkannya ke dalam database.

Java


import com.google.crypto.tink.Aead;
import java.security.GeneralSecurityException;
import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.SQLException;
import java.sql.Timestamp;
import java.util.Date;
import javax.sql.DataSource;

public class EncryptAndInsertData {

  public static void main(String[] args) throws GeneralSecurityException, SQLException {
    // Saving credentials in environment variables is convenient, but not secure - consider a more
    // secure solution such as Cloud Secret Manager to help keep secrets safe.
    String dbUser = System.getenv("DB_USER"); // e.g. "root", "mysql"
    String dbPass = System.getenv("DB_PASS"); // e.g. "mysupersecretpassword"
    String dbName = System.getenv("DB_NAME"); // e.g. "votes_db"
    String instanceConnectionName =
        System.getenv("INSTANCE_CONNECTION_NAME"); // e.g. "project-name:region:instance-name"
    String kmsUri = System.getenv("CLOUD_KMS_URI"); // e.g. "gcp-kms://projects/...path/to/key
    // Tink uses the "gcp-kms://" prefix for paths to keys stored in Google Cloud KMS. For more
    // info on creating a KMS key and getting its path, see
    // https://cloud.google.com/kms/docs/quickstart

    String team = "TABS";
    String tableName = "votes";
    String email = "hello@example.com";

    // Initialize database connection pool and create table if it does not exist
    // See CloudSqlConnectionPool.java for setup details
    DataSource pool =
        CloudSqlConnectionPool.createConnectionPool(dbUser, dbPass, dbName, instanceConnectionName);
    CloudSqlConnectionPool.createTable(pool, tableName);

    // Initialize envelope AEAD
    // See CloudKmsEnvelopeAead.java for setup details
    Aead envAead = CloudKmsEnvelopeAead.get(kmsUri);

    encryptAndInsertData(pool, envAead, tableName, team, email);
  }

  public static void encryptAndInsertData(
      DataSource pool, Aead envAead, String tableName, String team, String email)
      throws GeneralSecurityException, SQLException {

    try (Connection conn = pool.getConnection()) {
      String stmt =
          String.format(
              "INSERT INTO %s (team, time_cast, voter_email) VALUES (?, ?, ?);", tableName);
      try (PreparedStatement voteStmt = conn.prepareStatement(stmt); ) {
        voteStmt.setString(1, team);
        voteStmt.setTimestamp(2, new Timestamp(new Date().getTime()));

        // Use the envelope AEAD primitive to encrypt the email, using the team name as
        // associated data. This binds the encryption of the email to the team name, preventing
        // associating an encrypted email in one row with a team name in another row.
        byte[] encryptedEmail = envAead.encrypt(email.getBytes(), team.getBytes());
        voteStmt.setBytes(3, encryptedEmail);

        // Finally, execute the statement. If it fails, an error will be thrown.
        voteStmt.execute();
        System.out.println(String.format("Successfully inserted row into table %s", tableName));
      }
    }
  }
}

Python

import datetime
import logging
import os

import sqlalchemy
import tink

from .cloud_kms_env_aead import init_tink_env_aead
from .cloud_sql_connection_pool import init_db

logger = logging.getLogger(__name__)

def main() -> None:
    db_user = os.environ["DB_USER"]  # e.g. "root", "sqlserver"
    db_pass = os.environ["DB_PASS"]  # e.g. "mysupersecretpassword"
    db_name = os.environ["DB_NAME"]  # e.g. "votes_db"

    # Set if connecting using TCP:
    db_host = os.environ["DB_HOST"]  # e.g. "127.0.0.1"

    # Set if connecting using Unix sockets:
    db_socket_dir = os.environ.get("DB_SOCKET_DIR", "/cloudsql")

    instance_connection_name = os.environ["INSTANCE_CONNECTION_NAME"]
    # e.g. "project-name:region:instance-name"

    credentials = os.environ.get("GOOGLE_APPLICATION_CREDENTIALS", "")
    key_uri = "gcp-kms://" + os.environ["GCP_KMS_URI"]
    # e.g. "gcp-kms://projects/...path/to/key
    # Tink uses the "gcp-kms://" prefix for paths to keys stored in Google
    # Cloud KMS. For more info on creating a KMS key and getting its path, see
    # https://cloud.google.com/kms/docs/quickstart

    table_name = "votes"
    team = "TABS"
    email = "hello@example.com"

    env_aead = init_tink_env_aead(key_uri, credentials)
    db = init_db(
        db_user,
        db_pass,
        db_name,
        table_name,
        instance_connection_name,
        db_socket_dir,
        db_host,
    )

    encrypt_and_insert_data(db, env_aead, table_name, team, email)

def encrypt_and_insert_data(
    db: sqlalchemy.engine.base.Engine,
    env_aead: tink.aead.KmsEnvelopeAead,
    table_name: str,
    team: str,
    email: str,
) -> None:
    time_cast = datetime.datetime.now(tz=datetime.timezone.utc)
    # Use the envelope AEAD primitive to encrypt the email, using the team name as
    # associated data. Encryption with associated data ensures authenticity
    # (who the sender is) and integrity (the data has not been tampered with) of that
    # data, but not its secrecy. (see RFC 5116 for more info)
    encrypted_email = env_aead.encrypt(email.encode(), team.encode())
    # Verify that the team is one of the allowed options
    if team != "TABS" and team != "SPACES":
        logger.error(f"Invalid team specified: {team}")
        return

    # Preparing a statement before hand can help protect against injections.
    stmt = sqlalchemy.text(
        f"INSERT INTO {table_name} (time_cast, team, voter_email)"
        " VALUES (:time_cast, :team, CONVERT(varbinary(max), :voter_email, 0))"
    )

    # Using a with statement ensures that the connection is always released
    # back into the pool at the end of statement (even if an error occurs)
    with db.connect() as conn:
        conn.execute(stmt, time_cast=time_cast, team=team, voter_email=encrypted_email)
    print(f"Vote successfully cast for '{team}' at time {time_cast}!")

Membuat kueri database dan mendekripsi data yang disimpan.

Java


import com.google.crypto.tink.Aead;
import java.security.GeneralSecurityException;
import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.ResultSet;
import java.sql.SQLException;
import java.sql.Timestamp;
import javax.sql.DataSource;

public class QueryAndDecryptData {

  public static void main(String[] args) throws GeneralSecurityException, SQLException {
    // Saving credentials in environment variables is convenient, but not secure - consider a more
    // secure solution such as Cloud Secret Manager to help keep secrets safe.
    String dbUser = System.getenv("DB_USER"); // e.g. "root", "mysql"
    String dbPass = System.getenv("DB_PASS"); // e.g. "mysupersecretpassword"
    String dbName = System.getenv("DB_NAME"); // e.g. "votes_db"
    String instanceConnectionName =
        System.getenv("INSTANCE_CONNECTION_NAME"); // e.g. "project-name:region:instance-name"
    String kmsUri = System.getenv("CLOUD_KMS_URI"); // e.g. "gcp-kms://projects/...path/to/key
    // Tink uses the "gcp-kms://" prefix for paths to keys stored in Google Cloud KMS. For more
    // info on creating a KMS key and getting its path, see
    // https://cloud.google.com/kms/docs/quickstart

    String tableName = "votes123";

    // Initialize database connection pool and create table if it does not exist
    // See CloudSqlConnectionPool.java for setup details
    DataSource pool =
        CloudSqlConnectionPool.createConnectionPool(dbUser, dbPass, dbName, instanceConnectionName);
    CloudSqlConnectionPool.createTable(pool, tableName);

    // Initialize envelope AEAD
    // See CloudKmsEnvelopeAead.java for setup details
    Aead envAead = CloudKmsEnvelopeAead.get(kmsUri);

    // Insert row into table to test
    // See EncryptAndInsert.java for setup details
    EncryptAndInsertData.encryptAndInsertData(
        pool, envAead, tableName, "SPACES", "hello@example.com");

    queryAndDecryptData(pool, envAead, tableName);
  }

  public static void queryAndDecryptData(DataSource pool, Aead envAead, String tableName)
      throws GeneralSecurityException, SQLException {

    try (Connection conn = pool.getConnection()) {
      String stmt =
          String.format(
              "SELECT TOP(5) team, time_cast, voter_email FROM %s ORDER BY time_cast DESC;",
              tableName);
      try (PreparedStatement voteStmt = conn.prepareStatement(stmt); ) {
        ResultSet voteResults = voteStmt.executeQuery();

        System.out.println("Team\tTime Cast\tEmail");
        while (voteResults.next()) {
          String team = voteResults.getString(1);
          Timestamp timeCast = voteResults.getTimestamp(2);

          // Use the envelope AEAD primitive to encrypt the email, using the team name as
          // associated data. This binds the encryption of the email to the team name, preventing
          // associating an encrypted email in one row with a team name in another row.
          String email = new String(envAead.decrypt(voteResults.getBytes(3), team.getBytes()));

          System.out.println(String.format("%s\t%s\t%s", team, timeCast, email));
        }
      }
    }
  }
}

Python

import os

import sqlalchemy
import tink

from .cloud_kms_env_aead import init_tink_env_aead
from .cloud_sql_connection_pool import init_db
from .encrypt_and_insert_data import encrypt_and_insert_data

def main() -> None:
    db_user = os.environ["DB_USER"]  # e.g. "root", "sqlserver"
    db_pass = os.environ["DB_PASS"]  # e.g. "mysupersecretpassword"
    db_name = os.environ["DB_NAME"]  # e.g. "votes_db"

    # Set if connecting using TCP:
    db_host = os.environ["DB_HOST"]  # e.g. "127.0.0.1"

    # Set if connecting using Unix sockets:
    db_socket_dir = os.environ.get("DB_SOCKET_DIR", "/cloudsql")

    instance_connection_name = os.environ["INSTANCE_CONNECTION_NAME"]
    # e.g. "project-name:region:instance-name"

    credentials = os.environ.get("GOOGLE_APPLICATION_CREDENTIALS", "")
    key_uri = "gcp-kms://" + os.environ["GCP_KMS_URI"]
    # e.g. "gcp-kms://projects/...path/to/key
    # Tink uses the "gcp-kms://" prefix for paths to keys stored in Google
    # Cloud KMS. For more info on creating a KMS key and getting its path, see
    # https://cloud.google.com/kms/docs/quickstart

    table_name = "votes"
    team = "TABS"
    email = "hello@example.com"

    env_aead = init_tink_env_aead(key_uri, credentials)
    db = init_db(
        db_user,
        db_pass,
        db_name,
        table_name,
        instance_connection_name,
        db_socket_dir,
        db_host,
    )

    encrypt_and_insert_data(db, env_aead, table_name, team, email)
    query_and_decrypt_data(db, env_aead, table_name)

def query_and_decrypt_data(
    db: sqlalchemy.engine.base.Engine,
    env_aead: tink.aead.KmsEnvelopeAead,
    table_name: str,
) -> None:
    with db.connect() as conn:
        # Execute the query and fetch all results
        recent_votes = conn.execute(
            f"SELECT TOP(5) team, time_cast, voter_email FROM {table_name} "
            "ORDER BY time_cast DESC"
        ).fetchall()

        print("Team\tEmail\tTime Cast")

        for row in recent_votes:
            team = row[0]
            # Use the envelope AEAD primitive to decrypt the email, using the team name as
            # associated data. Encryption with associated data ensures authenticity
            # (who the sender is) and integrity (the data has not been tampered with) of that
            # data, but not its secrecy. (see RFC 5116 for more info)
            email = env_aead.decrypt(row[2], team).decode()
            time_cast = row[1]

            # Print recent votes
            print(f"{team}\t{email}\t{time_cast}")

Anda juga dapat membatasi akses pada tingkat instance atau database. ↩