1. Above the Clouds: A Berkeley View of Cloud Computing Annajiat Alim Rasel, 0409052002P Shimul Bala, 0412052105P Raquibul Bari, 0411052073P Annajiat Alim Rasel, 0409052002P Shimul Bala, 0412052105P Raquibul Bari, 0411052073P Distributed Computing Systems

2. UC Berkeley Above the Clouds A Berkeley View of Cloud Computing 2 Michael Armbrust et al.

3. Outline

4. Figure: Cloud Computing Shared utilization of Computing Resources With the abstraction of being utilized as a service rather than as a product Service model similar to utility services Computing Resources NetworksServers StorageApplications Services Cloud Computing 4 Back to Index

5. Cloud Computing (Contd.) NIST’s Final (16 th ) definition: Ubiquitous, convenient, on-demand network access Shared pool of configurable Computing Resources Resources can be rapidly provisioned and released Minimal management effort or service provider interaction NIST: National Institute of Standards and Technology 5 Back to Index

6. Deliver software as a service over the Internet Eliminating the need to install and run the application on the customer's own computers Simplifying maintenance and support Deliver a computing platform and/or solution stack as a service, Often consuming cloud infrastructure and sustaining cloud applications Deliver computer infrastructure- platform virtualization Environment – as a service, along with raw (block) storage and networking Cloud Computing Layers 6 Back to Index

7. Figure: Cloud Computing Layers 7 Back

8. Infrastructure operated solely for a single organization Shares infrastructure between several organizations from a specific community with common concerns Composition of two or more clouds Offering the benefits of multiple deployment models Deployment Models Figure: Cloud Computing Types Service provider makes resources, such as applications and storage, available to the general public over the Internet 8 Back to Index

9. Summary of Cloud Computing Utility Computing: Illusion of on-demand infinite resources Illusion of on-demand infinite resources Elimination of up-front cost Elimination of up-front cost Pay for usage Pay for usage Cloud Computing = SaaS + Utility Computing 9 Back to Index

10. Summary of Cloud Computing 10

11. Cloud Off The Ground Experience with  Economies of Scale, 1:5  1:7  Very Large Data Centers New Technology Trends, Apps Business Model  Standardized of Software Stack  Pervasive Internet  Usage based Billing Model  Advanced x86 Virtualization 11 Back to Index

12. Cloud Application Opportunities Mobile and Web Apps  Mashups, Real-time apps Extensions of Compute-Intensive Desktop Software  Matlab, Mathematica Batch Processing / MapReduce  The Washington Post 12

13. Classes of Utility Computing x86 Instruction Set Architecture (ISA) VM ByteCode based VM Application Framework VM EC2AzureAppEngineForce.com Lower Level More Control More Efforts Higher Level Less Efforts More Manageable 13 Back to Index

14. Unused resources Cloud Computing Economics Static Data Center Cloud Data Center Demand Capacity Time Resources Unused Time Resources 14 Pay only for what is used Back to Index

15. Static Data Center Demand Capacity Time Resources 15 Cloud Computing Economics Unused Over-Provisioning Risk - Underutilization

16. Resources Demand Capacity Time (days) 1 23 Cloud Computing Economics Under-Provisioning Risk  Lost Revenue  Lost Users 16

17. Economics of Cloud Providers Resource Cost in Medium DC Cost in Very Large DC Ratio Network$95 /Mbps/month$13 /Mbps/month7.1x Storage$2.20 /GB/month$0.40 /GB/month5.7x Administration≈140 servers/admin >1000 servers/admin 7.1x 17 Back to Index

18. Economics of Cloud Providers Oracle  sell licenses Amazon  utilize off-peak capacity Microsoft  sell.NET tools Google  reuse existing infrastructure 18

19. Cloud Adoption Challenges 19 Back to Index Challenge: Availability of Service Opportunity  Multiple Providers, Elasticity for DDoS Challenge: Availability of Service Opportunity  Multiple Providers, Elasticity for DDoS

20. Cloud Adoption Challenges 20 Back to Index Challenge: Data Lock-In Opportunity  Standardization of API  Surge Computing Challenge: Data Lock-In Opportunity  Standardization of API  Surge Computing

21. Cloud Adoption Challenges 21 Back to Index Challenge: Data Confidentiality & Auditability Opportunity  Geographical Data Storage  Encryption  VLANs  Firewalls Challenge: Data Confidentiality & Auditability Opportunity  Geographical Data Storage  Encryption  VLANs  Firewalls

22. Cloud Growth Challenges ChallengeOpportunity Data Transfer Bottlenecks FedEx-ing Disks, Data Backup/Archival Performance Unpredictability Improved VM support, Flash Memory, Scheduling VMs Scalable StorageInvent Scalable Store Bugs in Large Distributed Systems Invent Debugger on Distributed VMs Scaling Quickly Invent Machine Learning enabled Auto-Scaler; Snapshots 22

23. Policy and Business Challenges ChallengeOpportunity Reputation Fate Sharing Offer Reputation-Guarding Services like those for email Software Licensing Pay-for-use Licenses; Bulk use sales 23

24. Conclusion Short Term Implications: Startups and Prototyping One-Off Task: Washington Post, NY Times Research at Scale Cost Associativity  1 VM x 1000 hours = 1000 VMs x 1 hour 24 Back to Index

25. Conclusion Long Term Implications: 25 AreaImplication Application Software  Cloud & Client Parts  Disconnection Tolerance Infrastructure Software  Resource accounting  VM awareness Hardware Systems  Containers  Energy Proportionality

26. Reference 26 Back to Index Above the Clouds: A Berkeley View of Cloud Computing d1smfj0g31qzek.cloudfront.net/above_the_clouds.ppt www.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-28.html Unpublished work on Hybrid Clouds by Annajiat et al. National Institute of Standards and Technology's (NIST) www.nist.gov/itl/csd/cloud-102511.cfm Multiple citations from:  Amazon: EC2, AWS, EBS, S3, SimpleDB, CloudFront, Jeff  USENIX OSDI, ACM SIGOPS, ACM Queue, ACM SIGCOMM, IEEE MCSA, MapReduce, Cloudera Hadoop Training etc. Above the Clouds: A Berkeley View of Cloud Computing d1smfj0g31qzek.cloudfront.net/above_the_clouds.ppt www.eecs.berkeley.edu/Pubs/TechRpts/2009/EECS-2009-28.html Unpublished work on Hybrid Clouds by Annajiat et al. National Institute of Standards and Technology's (NIST) www.nist.gov/itl/csd/cloud-102511.cfm Multiple citations from:  Amazon: EC2, AWS, EBS, S3, SimpleDB, CloudFront, Jeff  USENIX OSDI, ACM SIGOPS, ACM Queue, ACM SIGCOMM, IEEE MCSA, MapReduce, Cloudera Hadoop Training etc.

27. Questions? 27 Back to Index