View of Cloud computing (CC)

• CC refers to both applications delivered as services over internet & hardware/software that provide those services
• DC hardware + software = cloud
  • public if made available
    • service being sold is utility computing
  • private is internal DCs of an org
• CC = sum of Saas & utility computing

New aspects

1. infinite computing resources on demand (don’t have to plan far ahead)
2. no upfront commitment
3. pay by use

Enablers

• construction / operation of large-scale commodity-computer DCs at low cost locations
  • decreased electricity cost
  • network bandwidth
  • operations / hardware

Classes

• application framework
  • AppEngine
• hybrid
  • azure
• hardware virtual machines
  • EC2

Obstacles & opportunities

1. Business continuity & service availability
   • only solution: multiple CC providers
2. Data lock-in
   • one solution is to deploy across multiple CC providers so that one down doesn’t end everything
   • keep local (surge computing)
3. Data confidentiality / auditability
   • primary security mechanism: virtualization
   • think about the virtualized capabilities
4. Data Transfer Bottlenecks
   • physical moment can be faster than bandwith bottlenecks
5. Performance Unpredictability
   • IO interference between VMs
     • might be decreased with flash memory
   • thread usage is a blackbox
6. Scalable Storage
   • how to combine this with durability, high availability, scaling
7. Bugs in Large-Scale Distributed Systems
   • Cannot be reproduced at a small scale
8. Scaling Quickly
   • reduce money
   • be environmentally friendly
   • need a fast & easy to use snapshot/restart tool
9. Reputation Fate Sharing
   • ban IPs
   • legal liability
10. Software Licensing
    • prices of proprietary vs open source

Conclusion

1. Need to scale up and down
2. Not bare metal -> VMs
3. Hardware systems developed at scale of a container (>= 1 dozen racks)