Computer Science vs Software Engineering

Sometimes the real world isn't black or white. Choosing the right tool requires context and trade offs. This was a lightning talk I gave for New Relic at Railsconf 2016.

Transcript

1. Computer Science vs Software Engineering Mike Perham Contributed Systems
3. A Tale of Locking and Giving Advice
4. @mperham: "Check out my locking API using Redis"
6. Context • He's both right and wrong. • What are you optimizing for?
7. Computer Science • Formal Correctness • Algorithmic Complexity • Black and White
8. Software Engineering • Welcome to the Real World™ • Cost of: • Maintenance • Operations • Getting Started • Performance, scalability, stability
9. Zookeeper • Pros • Formally correct Paxos consensus algorithm • Distributed, so 100% uptime (theoretically!)
10. Redis • Pros • Easy to get started • Fast and flexible • Widely available as SaaS • Cheap to operate
11. Zookeeper • Cons • More $$$ to operate • Minimum of 3 servers • No SaaS
12. Tony's View • Square • Deals with lots of $$$$$ • 100% uptime critical • JVM a positive • Already have a big, experienced ops team • "Use Zookeeper"
13. Mike's View • Sidekiq • Customers are from 1- to 1000-person companies • JVM can be unwelcome to some Rubyists • Many customers == NoOps • Higher operational costs == unhappy customers
14. Reality • No one is forced to use any API* • App developers decide what's important to them • Context is key
15. Pro Tips • Advice isn't helpful without considering context • Empathy improves your advice and makes you a better engineer
16. mperham {gmail,twitter,github}