When you are tackling many servers in the data center, saving a small percentage of servers would bring a significant return. Similarly, a seemingly small mistake can be very costly. We describe how we evaluate performance at scale, and also how it is different from optimization on a single system.

数据中心已经成为支撑大规模互联网服务的标准基础设施，随着数据中心的规模越来越大，节省数据中心一小部分的服务器也会带来非常可观的回报。与在单一系统上做性能优化相比，在大规模数据中心上进行性能分析有明显差别。

The emergence of large-scale software deployments in the data center has led to several challenges: (1) measuring software performance in the data center, and (2) evaluating the performance impact of software or hardware changes.

数据中心上的大规模软件部署会带来以下几个挑战：（1）如何在数据中心中衡量软件性能；（2）优化软件资源管理

We will highlight some challenges and solutions. Here is the session outline.

本次会议我们将分享一些为了解决这些挑战而使用的技术，详细描述如何组合常用工具来解决我们在数据中心中遇到的问题。主要内容包括：

1. Performance Efficiency Analysis In-depth look at resource utilization. Some metrics are misleading. （性能效率分析：在某些指标具有误导性的情况下，深入查看资源利用率）

2. Performance Evaluation at Scale: Trade-offs between accurate data and the large quantity of data Detect performance differences due to hardware changes, software changes, or in the presence of application interference. （大规模性能评估：由于硬件、软件改变或由于应用程序干扰而带来的检测性能差异，准确数据与大量数据之间的权衡）

Audience Take-Aways
1. Consider critical thinking in performance data collection and interpretation
2. Be wary of quick but wrong conclusions.