Cloud Computing: Amazon AWS and EC2
- 1. @TeamSkunkworks Cloud Computing: Building an Infrastructure with Amazon AWS and EC2 Chris Dollmont @MrCerulean
- 2. Agenda • Cloud Compu6ng Overview • Why? • How? • Discipline • Tools • Concept 2
- 3. Overview • What is the “Cloud”? • Where is it? • Who provides it? – Amazon – Rackspace – GoGrid – In house cloud 3
- 4. Overview • Amazon Services alphabet soup – EC2 (Elas6c Compute Cloud) – S3 (Simple Storage Service) – EBS (Elas6c Block Storage) – ELB (Elas6c Load Balancing) – SimpleDB – RDS (Rela6onal Database Service) – Route 53 (Elas6c DNS) – CloudWatch 4
- 5. Why? • Reduce hardware costs • Management • Ease of Deployment • Management • Scalability 5
- 6. Why not? • Reducing costs – bandwidth – cycles – storage – instances • Simplifying management – upfront investment 6
- 7. How? • Create an Amazon Web Services account – provide your credit card details • Launch an instance – Hundreds of publicly available instances – Linux, Windows, Solaris, BSD • Configure storage – S3 or EBS • Configure access 7
- 8. Discipline • Best prac6ces are cri6cal – Standard images – Standard tools – Define conven6ons • Support or not? – Paid support – Community support 8
- 9. Discipline • Moving to the cloud – Get your meatspace in order – Create your own AMI (image) – Use sane defaults • Test – And test – And test some more 9
- 10. Discipline • Loca6on, loca6on, loca6on – Be aware of latency – Amazon offers many loca6ons (regions) • US East • US West • EU (Ireland) • Asia/Pacific (Singapore) 10
- 11. Discipline • Decide on image availability – Reserved instances – Spot instances – On demand instances • Tie storage to an instance – Only when necessary • Create with transience in mind 11
- 12. Tools • Web based – Amazon’s AWS console • Free • Full featured – Rightscale • Not free • Excellent for full automa6on – SCALR • Not free • Excellent for full automa6on 12
- 13. Tools • Elas6fox – Graphical – Firefox based – 80% • Tim Kay’s aws tools – Command line – hep://www.6mkay.com/aws/ – scriptable (key!) 13
- 14. Tools • boto – Python based – hep://code.google.com/p/boto/ – command line – scriptable • mr.awsome – Python based – Uses boto – hep://pypi.python.org/pypi/mr.awsome – command line – scriptable 14
- 15. Tools • puppet – configura6on management • ZenOSS – system monitoring • flume – log aggrega6on • fabric – code deployment 15
- 16. Concept • Base AMI – Built on Debian Lenny – Stripped to the minimum – puppet client installed • Start new AMI – Assign a class – Authen6cate to puppet – Go get coffee 16
- 17. Concept • puppet – installs packages based on class – sets up configura6on – adds users and groups – sets up ssh auth – starts essen6al services 17
- 18. Concept • ZenOSS – Can autodiscover new hosts – Hosts can be grouped by func6on – Monitoring and stats gathering • flume – Centralizes logs – Allows for analysis 18
- 19. Ques%ons? @teamskunkworksvn, @MrCerulean #barcampsaigon #skunkworksvn
- 20. PSA 20/20/20 Fixing Powerpoint, one presenta6on at a 6me 20