Explore strategies for changing production environments without downtime in this 59-minute conference talk by Ryan Bergman from ChariotSolutions. Learn from real-world examples at John Deere, including moving a high-volume transactional monolith to AWS with minimal downtime, rewriting permission storage systems, and balancing storage vs. performance for a massive agronomic data pond. Discover techniques for cloud migration, breaking up monoliths, and implementing gradual changes through "crockpot migrations." Gain insights into handling growing data needs, API gateways, database migrations, and balancing machine learning requirements in large-scale systems. Understand the challenges and solutions involved in modernizing legacy systems and optimizing performance in cloud environments.
From Big Bangs to Crockpots: Strategies for Changing Production with No Downtime
Overview
Syllabus
Introduction
About Ryan Bergman
About John Deere
Growing Population
John Deere
Intelligent Solutions Group
War Stories
What Does ISG Do
Machine Sync
Data Collection
Data Structure
Tiled Format
Telematics
Big Bang Release
API Gateway
Central Monolith
Rest APIs
Data Centers
The Monolith
Problems Building Up
Amazon
Telnet
Oracle Golden Gate
Instant Near Instant
Sequences
Oracle Database
Firehose
Load Testing
Final Setup
Moving to Amazon
Breaking up monoliths
Cutting dependencies
User relationship tracking
Relationships get complicated
Our current implementation was broken up
We wanted to get off of the dependencies
The first swipe at this
The permissions table
Projections
Partner assignments
Crockpot migration
Crockpot analogy
Opting in partners
Getting rid of the old model
First stab at it
Bad points
Four distinct steps
Two paths
Older systems
Visual differences
Balancing
Migration
Machine Learning
Questions
Taught by
ChariotSolutions
