alembic不必使用数据库uri连接到数据库来运行升级。在你的 alembic\env.py 有这样一个函数：

def run_migrations_online():
    """Run migrations in 'online' mode.

    In this scenario we need to create an Engine
    and associate a connection with the context.

    """
    connectable = engine_from_config(
        config.get_section(config.config_ini_section),
        prefix='sqlalchemy.',
        poolclass=pool.NullPool)

    with connectable.connect() as connection:
        context.configure(
            connection=connection,
            target_metadata=target_metadata
        )

        with context.begin_transaction():
            context.run_migrations()

重要的是变量 connectable 是一个 engine alembic调用时的示例 connect() 在上面。
因此，您可以这样做（这不是测试，但我自己也做类似的事情）：

def run_migrations_online():
    """Run migrations in 'online' mode.

    In this scenario we need to create an Engine
    and associate a connection with the context.

    """

    sshtunnel.SSH_TIMEOUT = 5.0
    sshtunnel.TUNNEL_TIMEOUT = 5.0

    server =  sshtunnel.SSHTunnelForwarder(
        ('ssh.pythonanywhere.com', 22),
        ssh_password="mypassword",
        ssh_username="myusername",
        remote_bind_address=\
            (myname.mysql.pythonanywhere-services.com', 3306))
    server.start()

    connectable = create_engine(
        'mysql+mysqldb://mynameb:dbpassword@127.0.0.1:%s/dbname' % 
        server.local_bind_port
    )

    with connectable.connect() as connection:
        context.configure(
            connection=connection,
            target_metadata=target_metadata
        )

        with context.begin_transaction():
            context.run_migrations()

理想情况下，您应该将所有的连接逻辑放在一个可以从 alembic/env.py 在项目中只定义一次，然后就可以导入 engine 直接进入 env.py ，但你明白了。