Many business applications, i.e. online trading, e-commerce, and real-time databases, etc., demand the high standard of service. Failure to meet performance specifications may result in losing business or liability violations. Therefore, it becomes critical that the datacenter software must manage effectively resource sharing and contention, as more cores are integrated into multicore processors. Unfortunately, how to achieve high concurrency and performance has become increasingly challenging. In this talk, we will discuss several case study on resource sharing and contention. We will share our BKMs (Best Known Methods) on how to identify and optimize performance bottlenecks to deliver the extreme concurrency.
Audience will learn a systematic top-down approach to identify and root cause performance and scalability issues in complicated environment. In addition, the audience will learn how to develop optimized synchronization solution for datacenter software.