Distributed File Systems: Exercises

Background (SG Ch.17.1)
1. Define a service, a server, and a client.
2. What are some example file operations that might be provided by a file service.
3. What is the ideal image of a DFS?
4. What is the most important performance measure of a DFS?
5. What overheads does a DFS incur, compared to a conventional system?
Naming and Transparency (SG Ch.17.2)
1. What is naming in a file system?
2. What additional naming feature does a DFS have beyond a conventional system?
3. What is location transparency?
4. What is location independence?
5. What should a file name denote? Does static location independence do everything desired?
6. Give an example of a computer system that is diskless.
7. What feature does automount provide to a DFS?
8. How can total integration of component file systems be achieved?
9. Explain two disadvantages of the administrative complexity of NFS.
10. How does UNIX aggregate files?
11. What is a location-independant file ID?
Remote File Access (SG Ch.17.3)
1. What is the reason for using caching in a DFS?
2. Explain the basic concept of caching in a DFS.
3. How does a replacement policy keep the cache size bounded?
4. What are the smallest and largest possible grains of cached file data?
5. What are the main benefit and disadvantage of small grained caching?
6. How does network structure affect caching policy?
7. What is the benefit of disk cache over memory cache?
8. How does increased memory size affect caching policy? Why?
9. Describe the write-through policy for caching update.
10. What is the disadvantage of write-through?
11. What are two disadvantages to delayed write in caching?
12. Which caching update policy induces the least network traffic?
13. What are the two approaches to verifying the validity of a cache?
14. When is there a potential inconsistency when using the server initiated approach to consistency checking?
15. List three advantages of caching over remote access.
16. List three advantages remote access over caching.
Stateful vs Stateless Service (SG Ch.17.4)
1. Why is the performance of a stateful file service better than a stateless one?
2. How does a stateless file service avoid the need for state information?
3. Which approach copes with system crashes better?
4. What is the penalty for using a stateless service?
5. What is meant by idempotent file operations?
6. Does UNIX have a stateful or stateless file system?
7. Is NFS naturally implemented on UNIX?
File Replication (SG Ch.17.5)
1. In what kind of environment is file replication useful?
2. What is the basic requirement of a file replication scheme?
3. Should users be aware of replication?
4. What is the main problem with file replication?
5. If a user updates one copy of a file, what should happen transparently?
6. What tradeoff is there in file replication?
Example Systems (SG Ch.17.6)