tornado 使用 ThreadPoolExecutor 疑问

请教下: 假设我需要执行两次获取数据的操作, 分别需要耗时 5 秒和 10 秒, 也需要等待返回结果才继续处理, 要怎么弄?

class GetDataHandler(BaseHandler):
    executor = ThreadPoolExecutor(2)

    @tornado.web.asynchronous
    @tornado.gen.coroutine
    def get(self):
        t1 = self.executor.submit(self.get_something, self, 'url-1')
        t2 = self.executor.submit(self.get_something, self, 'url-2')
        data_1 = yield t1.result()
        data_2 = yield t2.result()
        ...
        self.write('OK')

    @run_on_executor
    def get_something(self, url):
    	print('start...')
        data = requests.get(url)
        return data

堵倒是不堵了, 也可以同时开始, 但返回结果就要 15 秒的样? 这个 yield 还是有点蒙