Study Questions

Chapter 17: Distributed File Systems

“Operating System Concepts 5/e”

Silberschatz, Galvin
CS 4210 Advanced Operating Systems

Fall 1999 · Georgia Tech/Computer Science · Hutto

 

 

1. What’s the difference between a server and a service?

 

2. What’s the difference between location transparency and location independence? Which is better? What are the advantages and disadvantages of each technique?

 

3. Why is caching very important for distributed file systems? Describe how caching issues arise at various levels in a DFS implementation (you can say a LOT about this question.)

 

4. NFS is a “stateless” file server. What does that mean? Why did the designers choose to build it as a stateless server? (What are the advantages?)

 

5. What does the phrase “consistency semantics” mean? Describe three different definitions of consistency used in distributed file systems.

 

5 Questions About Example Distributed File Systems

 

6. Unix United was essentially a “proof-of-concept” system with relatively poor performance. Why was the performance much worse than any of the other systems described?

 

7. NFS uses a “directory name lookup cache” to speed up “path name translation”. What is “path name translation”? Why is it slow in NFS?

 

8. Andrew scales to systems with thousands of workstations. What aspects of the Andrew design contribute to its scalability? Another way to answer this is: What is the fundamental difference between Andrew and NFS?

 

9. Sprite relates a DFS and virtual memory in a rather radical way. Explain.

 

10. Describe how Locus supports file replication.