Optimizing application performance - A first principles perspective

By the end of this talk, you'll know some mental models to think of application performance.

You'll know:

• Why running your program on twice the compute resources doesn't really make your program 2x faster
• How to design your programs to handle things concurrently
• How to apply the first principles approach in your own work to improve application performance and architecture

In this talk, I'll cover a brief journey of application performance and why it matters. I'll then talk about how you can think about improving the application performance if you think of the application as a blackbox.

I'll introduce Amhdal's law and its extension - Universal Scalability Law and talk about how it relates to application performance. We'll cover IO bound and CPU bound workloads and how you should think about performance improvements for these types of workloads.

Conclusion: After you attend this talk, you'll have a far better understanding and appreciation for the type of workloads you run and maintain in real world and you'll be able to apply appropriate performance improvement techniques to further optimize the programs you write.