Thread by @srhtcn

It is not possible to create a customer-centric product without culture and tooling. - @MarcJBrooker #reinvent

#DevOps is about a loop - @MarcJBrooker

@MarcJBrooker This loop can get broken if you refuse to analyze failures. Most of the time it is broken because we don't change that particular culture. - @MarcJBrooker #reinvent

"The risks that matter cannot be seen from the office" - @ri_cook #reinvent

@ri_cook "Looking at risks ahead of time is harder than analyzing incidents afterward. You can't know where you should invest from the office from a leadership position." - @MarcJBrooker #reinvent

@ri_cook @MarcJBrooker Everybody in Amazon has some operational responsibility. - @MarcJBrooker #reinvent

@ri_cook @MarcJBrooker "Humility is critical. There is nothing more humbling than coming together to analyze incidents and realizing that you made the wrong choice. You can't do change practices piece without humility." - @MarcJBrooker #reinvent #devops

Turn around this loop fast. Drive meachanisms to take back these learnings to the implementation. -@MarcJBrooker #reinvent #devops

@MarcJBrooker There may be reasons to have an ops team. But you have to close the loop. You have to have people in your organization getting incrementally smarter. #devops #reinvent @MarcJBrooker

"Chernobyl's red button made things worse." @MarcJBrooker #reinvent

@MarcJBrooker Great question: "How and why should the operators have compensated for design errors they did not know about? - Anatoly Dyatlov" @MarcJBrooker #reinvent

@MarcJBrooker Wicked learning environment: The things we learn don't match the environment. Our experience leads us to do the wrong thing. - @MarcJBrooker #reinvent

@MarcJBrooker We build your systems to be kind at Amazon. You don't want a big red button. @MarcJBrooker #reinvent

"Never settle for human mistake. Really easy to blame someone when we make a mistake." @MarcJBrooker #reinvent
Also some great recommendation on resillient engineering #devops

@MarcJBrooker When you add load to your system, latency will increase at some point. All you do is to delay the inevitable - @MarcJBrooker #reinvent

Some of the things we do at #AWS - @MarcJBrooker #reinvent

@MarcJBrooker This one is culturally difficult but critical: limit queue sizes.

@MarcJBrooker "Adding randomness adds resilience. As you retry in every on min or 15 min or whatever, that means you have spikes. The solution is: add Jitter" - @MarcJBrooker #reinvent

2nd how we do this at #AWS slide. Remember this!