Twitter's Quest for a Wholly Graal Runtime - Improving Java Performance

Explore Twitter's journey to optimize Java performance and reduce CPU utilization through the implementation of Graal runtime in this 41-minute conference talk. Dive into the potential for significant cost savings in large-scale distributed systems by generating better machine code. Learn about compact strings, finaglers, parallel garbage collection, and the comparison between C2 and C9 compilers. Discover the cost implications, potential savings, and the impact on requests per second. Gain insights into inlining, enterprise crawl, and the possibility of running 100% Java with Graal. Examine the challenges faced, including bugs and metadata issues, and understand why certain decisions were made. Consider the broader applications of Graal for other programming languages and how it can benefit large-scale distributed systems beyond Twitter.

Intro
Goal
Bugs
Compact Compact Strings
finagle
requests per second
parallelgc
growl
Metadata
C2 vs C9
Why X is Right Ratio
How Much Does It Cost
Can We Save Even More
We Are Happy
Whats Not Announced
Give It A Try
Inlining
Enterprise Crawl
Can Graal run 100 in Java
Other languages