Data Infrastructure as Code
如果无法正常显示，请先停止浏览器的去广告插件。
相关话题：
#zalando
1. Data Infrastructure
as Code
Building core data services in a small team
Michal Gancarski, Zalando SE
2. Introduction
Team Nucleo - Data Lake Core
❖
eight engineers, lead, producer and the product specialist
3. Introduction
The Infrastructure
❖
thousands of datasets, hundreds of users, tens of clusters
4. Introduction
Main Challenges
❖
❖
❖
❖
project overload
GDPR
compliance
job market
5. Introduction
Value Multipliers and Cost Savers
RoW - Return on Whatever
V - value
C - cost
6. (+V) Reach Out
Be Friends with Stakeholders
❖
❖
❖ compliance
security & governance
IT architecture
❖
❖ business intelligence
product analytics
7. (+V) Reach Out
Offer Multiple Support Channels
❖
❖
❖
❖
❖
team email
#datalake-users
#databricks-users
office hours
guest developers
8. (-C) Outsource & Empower
Watch Users Solving It For You
❖
users are your community
9. (-C) Outsource & Empower
Let Others Do Your Work
❖
innersourced Big Query pipeline & Presto proxy
10. (-C) Outsource & Empower
Lean on Your Vendors. Hard.
11. (+V) Automate and Generalize
Climb The Ladder of Automation
12. (+V) Automate and Generalize
Treat Use Cases as Future Services
❖
❖
pilot with one team
parametrize and turn into a service
13. (-C) Protect and Simplify
Build The Wall
14. (-C) Protect and Simplify
Maintain Identity and Vision
“I strongly believe that the future
prosperity of the American people
depends on how well
each data infrastructure team
understands what NOT to build.
#bigdata #yolo”
Abraham Lincoln in a letter to Congress, 1845
15. Thank you!
Michal Gancarski, Zalando SE