Your SlideShare is downloading. ×
0
What an Enterprise Can Learn from Netflix,
a Cloud Native Company
Yury Izrailevsky, VP Cloud and Platform Engineering, Net...
August 2008 Database Corruption
RDBMS
Scalability

Performance

Availability
Netflix Streaming Growth
•
•
•
•

5 billion quarterly streaming hours
40 million customers
41 countries
3 continents

100x...
Netflix Cross-regional Cloud Architecture
Scalability

Performance

Availability
Cloud Too Expensive?
87% cost reduction
per streaming start

Netflix data center
Cloud Efficiency Benefits
Streaming growth

1/4/2009

1/4/2010

1/4/2011

1/4/2012

1/4/2013

Cyclical daily streaming usa...
Scalability

Performance

Availability
A Truly Great Service… Has to Just Work!

Availability goal: 99.99%
30 secs/week
at peak traffic
10/13/2013

9/22/2013

9/1/2013

8/11/2013

7/21/2013

6/30/2013

6/9/2013

5/19/2013

4/28/2013

4/7/2013

3/17/2013

2/2...
Netflix Cloud Journey: Tough Decisions
•
•
•
•

System rearchitecture
New security model
New operational model
Organizatio...
Old Architecture: Consolidated Java App

…

…
Javaweb

Javaweb

Javaweb
Cloud Native Service-based Architecture

15
Cascading Failures

API

Instant
Queue

Simple DB
16
Cascading Failures

…

X
99% availability

99%

99% availability

500

99% availability

= 0.657%
17
Cloud Native: Strategies to Improve Availability

Graceful
degradation

Redundancy
Cloud Native: Graceful Degradation
Cloud Native: Redundancy

Zone
A

Zone
B

Zone
C

Redundancy across
Availability Zones
Cloud Native Persistence

RDBMS

Relational NoSQL
distributed databases
Testing Fault Tolerance: Simian Army

Chaos Monkey

Latency Monkey

Chaos Gorilla
Open Source Portal at http://netflix.github.com
Cloud Native Operations: DevOps
Netflix data center

Central NOC team
coordinates biweekly releases

Dev teams push produc...
AMI-Based Cloud Deployments

Old
code

Bake new AMI for
each app deployment

New
code

Red-black
deployments
Evolving a Cloud Native Organization
Data center
IT-Ops manages
budget, capacity

Self-service provisioning by dev
teams; ...
Cloud Pilot Project: Jobs Page
Building a Great Streaming Product

28
Please give us your feedback on this
presentation

ENT203
As a thank you, we will select prize
winners daily for completed...
What an Enterprise Can Learn from Netflix, a Cloud-native Company (ENT203) | AWS re:Invent 2013
Upcoming SlideShare
Loading in...5
×

What an Enterprise Can Learn from Netflix, a Cloud-native Company (ENT203) | AWS re:Invent 2013

745

Published on

In moving its streaming product to the cloud, Netflix has been able to realize tremendous benefits in scalability, performance, and availability. The biggest benefit came from moving to a service-based architecture, which allowed engineering teams to accelerate their development cycle and innovate more quickly. However, cloud migration was a substantial effort. We mobilized resources across the company over several years, reorganized our engineering and operations teams, developed new security policies, migrated to the DevOps operations model, and even embraced a new product architecture. In this talk, we trace the evolution of the Netflix cloud model, both the successes and the challenges, and present them in a way that’s maximally useful to enterprises considering making the move to the cloud.

Published in: Technology, Business
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
745
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
41
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide

Transcript of "What an Enterprise Can Learn from Netflix, a Cloud-native Company (ENT203) | AWS re:Invent 2013"

  1. 1. What an Enterprise Can Learn from Netflix, a Cloud Native Company Yury Izrailevsky, VP Cloud and Platform Engineering, Netflix November 14, 2013 © 2013 Amazon.com, Inc. and its affiliates. All rights reserved. May not be copied, modified, or distributed in whole or in part without the express consent of Amazon.com, Inc.
  2. 2. August 2008 Database Corruption RDBMS
  3. 3. Scalability Performance Availability
  4. 4. Netflix Streaming Growth • • • • 5 billion quarterly streaming hours 40 million customers 41 countries 3 continents 100x growth since 2009
  5. 5. Netflix Cross-regional Cloud Architecture
  6. 6. Scalability Performance Availability
  7. 7. Cloud Too Expensive? 87% cost reduction per streaming start Netflix data center
  8. 8. Cloud Efficiency Benefits Streaming growth 1/4/2009 1/4/2010 1/4/2011 1/4/2012 1/4/2013 Cyclical daily streaming usage Economy of scale Elasticity
  9. 9. Scalability Performance Availability
  10. 10. A Truly Great Service… Has to Just Work! Availability goal: 99.99% 30 secs/week at peak traffic
  11. 11. 10/13/2013 9/22/2013 9/1/2013 8/11/2013 7/21/2013 6/30/2013 6/9/2013 5/19/2013 4/28/2013 4/7/2013 3/17/2013 2/24/2013 2/3/2013 1/13/2013 12/23/2012 12/2/2012 11/11/2012 10/21/2012 9/30/2012 9/9/2012 8/19/2012 7/29/2012 7/8/2012 6/17/2012 5/27/2012 5/6/2012 4/15/2012 3/25/2012 3/4/2012 2/12/2012 1/22/2012 1/1/2012 12/11/2011 11/20/2011 10/30/2011 10/9/2011 9/18/2011 8/28/2011 8/7/2011 7/17/2011 Weekly Streaming Availability (13wkMA) 12/24/2012 Elastic Load Balancing outage Using AWS redundancy to build highly fault-tolerant architecture
  12. 12. Netflix Cloud Journey: Tough Decisions • • • • System rearchitecture New security model New operational model Organizational changes
  13. 13. Old Architecture: Consolidated Java App … … Javaweb Javaweb Javaweb
  14. 14. Cloud Native Service-based Architecture 15
  15. 15. Cascading Failures API Instant Queue Simple DB 16
  16. 16. Cascading Failures … X 99% availability 99% 99% availability 500 99% availability = 0.657% 17
  17. 17. Cloud Native: Strategies to Improve Availability Graceful degradation Redundancy
  18. 18. Cloud Native: Graceful Degradation
  19. 19. Cloud Native: Redundancy Zone A Zone B Zone C Redundancy across Availability Zones
  20. 20. Cloud Native Persistence RDBMS Relational NoSQL distributed databases
  21. 21. Testing Fault Tolerance: Simian Army Chaos Monkey Latency Monkey Chaos Gorilla
  22. 22. Open Source Portal at http://netflix.github.com
  23. 23. Cloud Native Operations: DevOps Netflix data center Central NOC team coordinates biweekly releases Dev teams push production changes on own schedule; no central coordination
  24. 24. AMI-Based Cloud Deployments Old code Bake new AMI for each app deployment New code Red-black deployments
  25. 25. Evolving a Cloud Native Organization Data center IT-Ops manages budget, capacity Self-service provisioning by dev teams; visibility through tools Coordinated releases via centralized NOC Distributed DevOps; SREs build tools, share best practices Oracle DBAs manage several databases Java, DevOps engineers support dozens of Cassandra clusters Data science: analysts write SQL queries Hadoop engineers build ETL using PIG/Python
  26. 26. Cloud Pilot Project: Jobs Page
  27. 27. Building a Great Streaming Product 28
  28. 28. Please give us your feedback on this presentation ENT203 As a thank you, we will select prize winners daily for completed surveys!
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.

×