Go at Mainframe Scale

Explore the journey of implementing Go at mainframe scale in this GopherCon 2021 talk by Kaylyn Gibilterra. Discover how Capital One created a distributed system in Go to process account data from modern cores and legacy mainframes, handling peak traffic of 2 million tps and 5Gbps while storing over 20TB across multiple AWS regions. Learn about the challenges and successes of transitioning to Go services, including library creation, scaling issues, and team ramp-up. Gain insights into mainframe history, communication challenges, and technical hurdles faced during the implementation. Understand the importance of interfaces, typing, and library development in Go. Explore testing best practices, concurrency findings, and the availability of IBM Open Enterprise SDK for Go. Get a practical guide on how any company can transition to writing Go services from any language, leveraging Go's simplicity and scalability to achieve linear team growth and exponential service scaling.

Intro
Go at Mainframe Scale • Part 1: Mainframe Scale
The Mainframe Origin Story • Howard Alken created the first Mainframe at IBM, called the Harvard Mark I in 1944
Scale Today
Who is still using mainframes Communications
How we want to mainframe merges to work
How they actually work
Technical Challenges
6 months in
Failed attempts at replacing mainframes
My Mainframe Scale Leamings
Mainframe Adjacent Systems
Gopher phase
Experience across an organization
Language Background
Story: Grace Hopper
MVT Objectives
Interfaces
Typing
Library Development
Why manually test? Case Study: Systematics
Testing Best Practices
The Poor (wo)man's Code Review
Banking Systems
Upsetting Concurrency Findings Where we need it: • Saving into databases (sometimes) • Making parallel calls to dependencies Where don't we need it! . Most of our systems
Total Volume
IBM Open Enterprise SDK for Go is now available!