P4L2 - Distributed File Systems Flashcards

1
Q

T/F: Modern operating systems export a high-level filesystem interface to abstract all of the different types of storage devices present on a machine

A

True

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is a Distributed Filesystem

A

Environments that involve multiple machines for the delivery of the filesystem service

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is the replicated system DFS model

A

All files are replicated and available on every sever machine

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What are the benefits of the replicated system DFS model

A
  1. Fault Tolerant

2. Highly Available

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is the partitioned DFS model

A

Each server holds only some subset of the total files

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What are the benefits of the partitioned DFS model?

A
  1. Scalable
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

T/F: It’s uncommon to see a DFS that utilizes both partitioning and replication

A

False

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is the 3rd alternative to partitioned/replicated DFS models?

A

Files stored and served from all machines. All systems are peers

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

A _____ server keeps no notion of which files/blocks are being accessed, which operations are being performed, how many clients are accessing how many files

A

stateless

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

In a stateless server, every request must be _______

A

self-contained

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

T/F: Stateless servers can be used with models that rely on caching?

A

False! Without state, we CANNOT achieve consistency management

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What are some benefits of stateless server?

A
  1. Since there is no state on the server, there is no CPU/memory utilization required to manage that state
  2. Design is very resilient
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

A _____ server maintains information about the clients in the system, which files are being accessed, which types of accesses are being performed, which clients have a file cached, which clients have read/written the file

A

stateful

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

T/F: Stateful servers can be used with caching models

A

True

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Describe session semantics

A

Client writes back whatever data was modified on CLOSE. Whenever a client needs to open a file, the cache is skipped

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

T/F: Session semantics allow a client to be potentially reading from a stale file?

A

True!

17
Q

In what scenario is session semantics a poor choice?

A

Situations were clients want to concurrently share a file

18
Q

Describe periodic update

A

In order to avoid long periods of inconsistency, the client may write back changes periodically to the server

19
Q

T/F: It is not uncommon to adopt one type of semantics of files and one for directories

A

True

20
Q

In a ____ _____ ____ clients access files across the network

A

Networking file system (NFS)

21
Q

What is responsible for determining if a file belongs to the local file system or whether the request needs to be pushed to the NFS client?

A

The VFS layer (Virtual File system)

22
Q

The NFS client interacts with the NFS server via ____

A

RPC

23
Q

When an open request comes to an NFS server, it will create a ____ ____

A

file handle

24
Q

What is the purpose of a file handle?

A

It is used in NFS to contain information about the server machine as well as information about a file

25
Q

NFSv3 is ____ while NFSv4 is ______

A

stateless

stateful

26
Q

For files that are not accessed concurrently what type of semantics does NFS employ?

A

Session semantics

27
Q

T/F: NFS does not support periodic updates

A

False!

28
Q

According to “Caching in the Sprite Network File System” what percent of file access are writes?

A

33%

29
Q

According to “Caching in the Sprite Network File System” what percent of files were open less than half a second

A

75%

30
Q

According to “Caching in the Sprite Network File System” what percent of new data was deleted within 30 seconds

A

20-30%

31
Q

List the primary design decisions in implementing a DFS

A
  • Driver (client or server)
  • Redundancy/Replication
  • Update Semantics
  • Tracking State
  • Caching
32
Q

What are the positives of the Upload/Download model?

A
  • A client downloads the entire file, so it can operate on it locally.
33
Q

What are the negatives of the Upload/Download model?

A
  • The client has to download the entire file even for small file modifications
  • The server doesn’t know when it will get the file back or what state it will be in.
34
Q

What are the positives of the Remote File Access model?

A
  • The file remains on the server, so the server has full control and knowledge over the files. State control is easier.
35
Q

What are the negatives of the Remote File Access model?

A
  • Network latency - even when repeatedly reading

* Server will be overloaded more quickly/Limits scalability

36
Q

What are the negatives of a Stateful server?

A

The server must keep track of the state of all files, which means:

  • More complicated error recovery
  • Overheads to maintain the state.
37
Q

Where can the files or file blocks be cached in a single file server with many clients?

A
  • In client memory
  • On client storage device
  • In buffer cache in memory on server