Distributed File

8/2/2019 Distributed File

1/15

Distributed System Based File

System


2/15

Syncronization

Semantic File Sharing

File Locking

Sharing File in Coda


3/15

Semantics of File Sharing (1)

a) On a single processor, when a readfollows a write, the value returnedby the readis the value justwritten.

b) In a distributed system withcaching, obsolete values may bereturned.


4/15

Semantics of File Sharing (2)

Four ways of dealing with the shared files in a distributed

system.

Method Comment

UNIX semantics Every operation on a file is instantly visible to all processes

Session semantics No changes are visible to other processes until the file is closed

Immutable files No updates are possible; simplifies sharing and replication

Transaction All changes occur atomically


5/15

File Locking in NFS (1)

NFS version 4 operations related to file locking.

Operation Description

Lock Creates a lock for a range of bytes

Lockt Test whether a conflicting lock has been granted

Locku Remove a lock from a range of bytes

Renew Renew the leas on a specified lock


6/15

File Locking in NFS (2)

The result of an open operation with share reservations in NFS.a) When the client requests shared access given the current denial state.

b) When the client requests a denial state given the current file access state.

Current file denial state

NONE READ WRITE BOTH

READ Succeed Fail Succeed Succeed

WRITE Succeed Succeed Fail Succeed

BOTH Succeed Succeed Succeed Fail

(a)

Requested file denial state

NONE READ WRITE BOTH

READ Succeed Fail Succeed Succeed

WRITE Succeed Succeed Fail Succeed

BOTH Succeed Succeed Succeed Fail

(b)

Request

access

Current

access

state


7/15

Sharing File in Coda

The transactional behavior in sharing file in Coda


8/15

Concistency And Replication

Client Side Caching Caching in NFS

Client Side Caching in Coda

Client Side Caching for portable device

Server Side Replication Server Replication in Coda

Replication in Peer-to-Peer File Systems

Unstructured Peer-to-Peer Systems Structured Peer-to-Peer Systems

File System Replication in Grid System


9/15

Client Caching (1)

Client-side caching in NFS.


10/15

Client Caching (2)

Using the NFS version 4 callback mechanism to recall filedelegation.


11/15

Client Caching (3)

The use local copies when opening session in Coda


12/15

Client Caching (4)

Client Side Caching in Portable Device Storage devices not permanently connect to the system.

Case: upload/download model is used for maintaining files onportable storage devices.

Matters can be simplified if the storage device is viewed as partof a distributed file system.Tolia et al. (2004) propose to take a very simple approach by

storing locally a cryptographic hash of the data contained in files.


13/15

Server Side Replication

Server Replication in Coda

Two clients with a different AVSG for the same replicated file


14/15

Replication in Peer-to-Peer File Systems

Unstructured

Looking up data boils down to searching that data in the network.

A node broadcast a search query to its neighbors, from where the querymay be forwarded, and so on

Replication happens naturally when users download files from others andsubsequently make them available to the community.

Structured

Deployed to balance the load between the nodes

To simply replicate a file along the path that a query has followed fromsource to destination


15/15

File System Replication in Grid System

A key observation is that in many Grid applicationsdata are read only. Data often produce but rarelyupdate. So data replication can be applied abundance.

Replication in Grid systems mainly evolves aroundthe problem of locating the best sources to copy datafrom.

The problem is solved by special replica location

services. Globus toolkit

Distributed File

Documents

Transcript of Distributed File