Difference between revisions of "DistOS 2014W Midterm Review"

From Soma-notes
Jump to navigation Jump to search
Line 21: Line 21:
* The architecture of distributed OSs typically are some mix of peer-to-peer and hierarchies (e.g. master/slave) in the arrangement of servers.  Give an example of each and explain why the choice was made (a hierarchy versus peer-to-peer).
* The architecture of distributed OSs typically are some mix of peer-to-peer and hierarchies (e.g. master/slave) in the arrangement of servers.  Give an example of each and explain why the choice was made (a hierarchy versus peer-to-peer).
* "UNIX compatibility is increasingly sacrificed in order to improve scalability."  Argue for or against.
* "UNIX compatibility is increasingly sacrificed in order to improve scalability."  Argue for or against.
* Ceph is the solution to what problem?  How does it solve that problem?  (Choose at least one problem to discuss.)
* Describe three strategies used to increase scalability in distributed operating systems.  Give and explain specific examples of each.

Revision as of 12:01, 25 February 2014

Themes

  • Scaling
  • caching
  • consistency
  • specialized versus general solutions
  • OS abstractions
  • paths not taken
  • network transparency (making remote look like local)
  • reliability
    • fault tolerance
    • redundancy
    • availability
  • resource sharing
  • information sharing
  • communication
  • collaboration
  • peer-to-peer versus hierarchy
  • latency versus bandwidth

Questions

  • The architecture of distributed OSs typically are some mix of peer-to-peer and hierarchies (e.g. master/slave) in the arrangement of servers. Give an example of each and explain why the choice was made (a hierarchy versus peer-to-peer).
  • "UNIX compatibility is increasingly sacrificed in order to improve scalability." Argue for or against.
  • Ceph is the solution to what problem? How does it solve that problem? (Choose at least one problem to discuss.)
  • Describe three strategies used to increase scalability in distributed operating systems. Give and explain specific examples of each.