A presentation at DevOps Tallinn 2018 in in Tallinn, Estonia by Ryan Townsend
Don’t Panic! How to launch a large-scale website confidently and successfully Photo by SpaceX on Unsplash DevOps Tallinn 2018
Who am I? @ryantownsend Ryan Townsend, CTO
Relaunched May 2017
“Just use auto-scaling and forget about it” Kris Quigley – Lead Developer @ SHIFT (sarcasm)
Timeline Development Pre-launch Launch ! Post-launch
• Functional Testing • Deployment Pipelines • Configuration & Implementation
Development http://www.spacex.com/media-gallery/detail/149431/9391
Keep Things Simple
Limit Project Scope
New Problem or New Technology
“A l m o s t a l l t h e c a s e s w h e r e I ' v e h e a r d o f a
system that was built as a microservice system
from scratch, it has ended up in serious trouble.”
– Martin Fowler, ThoughtWorks CTO
Clear Decoupling
Admin Panel API Website
Use Boring Mature Technology
Load Testing
Don’t wait until the end
It’s A LOT harder than people let on
• Use real metrics and logged user behaviour • Use a wide variety of metrics, not just traffic • Post-test validate the metrics at source
Assume user behaviour will change
Stress Test
Web Performance Testing
Remember: it’s not just for you!
Caching
Client CDN Application Database
Write-through caches
Start small… low TTLs
Front-end – static assets & redirects
Higher hit ratios = less traffic hitting our servers
Feature Toggles
Ideal Fallback Off On
On Ideal Fallback Off
• Built into your application • Content Delivery Network • A/B testing tool
Circuit Breakers
Ideal Fallback Open Error Closed
Ideal Fallback Open Error Closed
Ideal Fallback Open Error Closed
Pre-launch Preparations https://www.flickr.com/photos/spacex/31450835954/
Communication
• Build a trusting relationship with stakeholders • Understand their metrics • Get their perspective • Determine authority
Visibility
• System monitoring – infrastructure & client-side
• Client / stakeholder dashboards & reporting – see what they see
• Customer engagement – social media, customer support
• Instant access to logs – filterable, searchable
Above shows how New Relic tracked a 3rd party script harming site performance but the server-side was fine.
Roleplay
• What could go wrong? • Who would you escalate to? • How would you solve? • What people do you need access to? • What systems do you need access to?
Traffic Reduction
• Avoid scheduling big campaigns • Paid advertising is easy to turn off • Reduce offering
Launch Day https://unsplash.com/photos/yJv97tE7GDM
Scale-up
“Big Bang” vs Canary Release
Feature Toggles: Off
Keep Calm and Carry On
• Expect issues • Keep a level-head • Remain professional • You’re an expert – you’ve got this "
Post-launch https://unsplash.com/photos/-p-KCm6xB9I
Continue Building Confidence
• Gather actual real metrics & usage patterns • Revisit your load tests and re-assess • Re-run load tests for future releases • Ship some safe releases • Ship small releases, often
Since Launch https://unsplash.com/photos/MEW1f-yu2KI
Optimising Caching
Strong Migrations
Started working towards micro macro-services
Event Sourcing
Static Site Generation
Communication is Paramount
Thank you
@ryantownsend
Launch-days can be stressful, particularly when the project’s key stakeholders are keeping an extra keen eye on how things are going. Back in May 2017, my team felt this pressure all too well as we relaunched the website of billion-pound UK retailer, Matalan, on our own ecommerce platform, SHIFT. As exciting as it was, there were tense moments as we opened the floodgates and let an unprecedented amount of traffic begin hitting the site in an instant.
In this talk, I’ll step through some of the technical and non-technical lessons that we learned in the process, and leave you better prepared and more confident when doing the same.
Here’s what was said about this presentation on social media.
for free. You
can too.