Presentation: Seamlessly Migrating To Serverless with 80-Million Users

Track: Finding the Serverless Sweetspot

Location: Broadway Ballroom South, 6th fl.

Duration: 11:50am - 12:40pm

Day of week:

Slides: Download Slides

Level: Intermediate

Persona: Architect

What You’ll Learn

  • Understand the philosophy of how trends like serverless and GraphQL are allowing business to reduce things down to its theoretical minimum computing?

  • Gain an intuition and understanding how serverless can be used in your environment.

  • Understand better how to reason about the tradeoffs, costs, and devops considerations involved with serverless environments.

Abstract

A deep dive into how serverless won over traditional approaches to web stacks. How it impacted DevOps, costs, and what we look forward to in the future.

Question: 

Is your talk a story of a migration to a serverless architecture through a devops lens or maybe are you talking about practices and patterns for serverless? What's the angle that you're taking with your talk?

Answer: 

I'm starting with just building a little bit of a foundation for what I feel devops actually is. There's a lot of charts about like the process project creation, packaging, releasing, and monitoring in a serverless world. Then start talking about devops for us.

The way it's heading is to present what 2015 looked like and this is the sort of tooling we're using now. Then dig into a few devops topics with code examples. I'll cover things like testing, configuration, and specifics about how we're using tooling.

Question: 

What I'm hearing is that serverless is a part of it, but the big theme is devops. Is that right?

Answer: 

Yeah. It's basically painting a picture of serverless devops today. We placed a bet on serverless. These are the things that happened along the way and where we wound up.

Question: 

What's the big message you're hoping to get out?

Answer: 

I plan to discuss the philosophy of serverless. Things like how a lot of people look at the ecosystem and compare sort of serverfull devops to serverless devops. I'll be making an argument that it's not really an evolution as much as a total paradigm shift (with examples) when it comes to serverless. I'll get in depth with testing, configuration, deployment, and monitoring. Then I'll wrap that up with a cost analysis. The cost analysis part is because a lot of people ask what's your EC2 bill now versus your Lambda API Gateway bill before? Most people want to see those comparisons. When I offer the context I mentioned, there's more of an informed conversation that we can have about how you can measure the cost.

Question: 

You mentioned theoretical minimum compute earlier when we were talking. What does that mean?

Answer: 

Theoretical Minimum Compute is the minimum amount of resources you need to consume to get the job done. That’s really the paradigm shift we’re talking about with serverless. It's a big part of what you can achieve with serverless. GraphQL is another technology that allows you to focus on the minimum about of resources to get the work done.

Speaker: Tyler Love

CTO @bustle

Tyler Love is CTO and co-founder of Bustle, where he leads the 20-person product team, protects their time and focus, and facilitates their ideas. He oversees a technical stack that serves billions of page views a month to 80 million unique visitors, built on experience helping to scale Tumblr from 500 million to 15 billion monthly page views, after tackling similar problems in the early years of Bleacher Report.

Find Tyler Love at

Similar Talks

Self-Selection for Resilience and Better Culture

Qcon

Agile/DevOps Trainer & Founder of Agile Play Consulting, LLC

Dana Pylayeva

CockroachDB: Architecture of a Geo-Distributed SQL Database

Qcon

CockroachDB maintainer, Co-founder & CTO @CockroachDB

Peter Mattis

Breaking Hierarchy - How Spotify Enables Engineer Decision Making

Qcon

Senior Engineering Manager, Data and Machine Learning Infrastructure @Spotify

Kristian Lindwall

The State of Serverless Computing

Qcon

CS PhD student at RISELab, UC Berkeley

Chenggang Wu

Video Streaming at Scale

Qcon

IBM Distinguished Engineer, CTO Watson Media Cognitive Solutions @IBM

Lysa Banks

Machine-to-Machine Interfaces

Qcon

Sr. Consultant, AppDev @awscloud

Ari Lerner