Global Memory Management
------------------------

1) What is the goal of the global memory management?  How does the
   memory management policy achieve this goal?

2) Assume 4 nodes (N1 through N4) each with 16 physical page frames.
   Initially none of the nodes have any valid pages in their physical
   memory.  N1 and N2 each make a sequence of page accesses to 32 distinct
   pages all from the disk.   N3 and N4 are idle.  What is the state of
   the cluster (i.e. local and global caches at each node) at the
   end of the above set of accesses?  

3) How is the minimum age of a page (cluster wide) determined?  Explain
   the algorithm to determine the MinAge.

4) Why is node failure not an issue in global memory management?

5) Explain the functionality of each of the data structures (PFD, GCD, POD)
   in the global memory management scheme.

6) In the actual implementation, how does each node collect the age
   information for its pages?

7) With reference to Table 1, explain why GCD processing time is 
   considerably less when the page is non-shared.

8) Summarize the conditions under which GMS can hope to significantly
   improve the performance for file accesses compared to vanilla NFS.

   
TreadMarks
----------

1) Distinguish between memory coherence and memory consistency model.

2) Explain the following terms with code samples to illustrate the point:
   a) data race
   b) false sharing

3) Explain the difference between sequential consistency and release
   consistency with code samples.

4) Explain the difference between eager and lazy release consistency
   with code samples.  What are the pros and cons of each approach?

5) How does the TreadMarks DSM solve the false sharing problem?
   Explain the protocol details.

6) Given that TreadMarks is a user level DSM package, What support is expected 
   from the Operating System for implementing it?

7) Develop the data structures and pseudo code for the TreadMarks 
   DSM.  Take into account the distributed lock manager and the 
   protocol actions needed commensurate with the memory consistency model
   and the coherence mechanism of TreadMarks.

8) What are the sources of overhead in the implementation of TreadMarks?

9) What is the need for garbage collection in the TreadMarks implementation?
   Explain.

Serverless Network File System
------------------------------

1) What are the technological enablers justifying a serverless NFS?
   
2) xFS builds on three prior technologies: RAID, Zebra LFS, and multiprocessor
   cache consistency.  Discuss the limitations in each of these technologies
   with respect to realizing a scalable serverless NFS.

3) In a centralized file system, the server performs the functions of 
   managing the data blocks, metadata for the files, server-side
   file cache, and consistency of datablocks of files cached by
   multiple clients.  Discuss how these functions are carried out in
   xFS. 

4) What is the advantage of the stripe group concept used in xFS?

5) Explain the idea of log cleaning and why it is needed.  Discuss the
   principle and details of the distributed log cleaners employed in xFS.

6) The paper claims: "For instance, in a 32 node xFS system with 32 clients,
   each client receives nearly as much read or write bandwidth as it would
   see if it were the only active client."
   Explain this statement.