Distributed file system

Distributed file system

• 1.
  Distributed File System
• 2.
  Introduction  File –collection of data with user view(structure) and a physical view (blocks)  DFS – distributed storage and users – provides transparency of location, access and migration of files – uses replicas for efficiency and fault tolerance  Enables programs to store and access remote files as local ones  Two types  Andrew File system (AFS)  Sun Network File system (NFS)  File systems are abstractions that enables users to read, manipulate and organize dat. Data is stored in units known as files in a hierarchical tree – nodes are directories
• 3.
   Enables uniformview – independent of underlying storage devices (floppy to hard drives and flash memory cards)  Goals:  Network transparency – access rremote files  High availability – easy to access files irrespective of their physical location
• 4.
  Characteristics  Organization, storage,retrieval, naming sharing and protection of files  File – Data + attributes (length, timestamp, file type)  Directory – provides mapping from names to internal file identifier  Modules  Directory module – File names to File IDS  File module – File IDS to particular files  Access control module – checks permission for file operation  Block module – Access and allocates disk blocks  Device module – Disk IO and buffering
• 6.
  Requirements 1. Transparency  Structure Access  Naming  Replication 2. User mobility 3. Performance 4. Scalability 5. High Availability 6. High reliability 7. Security
• 7.
  File Service Architecture Clear separation of main concerns providing access to files is obtained by structuring the file service as three components – Flat file service, directory file service, client module
• 8.
   Flat fileservice  Implementation of operations on the content of the file  UFID – Unique File Identifier – refer to files in all requests for flat file operations  UFID – long sequence of bits chosen  Create file – new UFID will be generated and returns UFID
• 9.
   Directory service Mapping between text names for the files and their UFIDs  Get UFID by quoting the text name to directory service  It supports functions needed to generate directories, add files to directories
• 10.
   Client module Runs on each computer and provides integrated service as a single API to application programs.  Holds information about the network locations of flat file, directory server processes  Achieves better performance through implementation of cache of recently used file blocks
• 11.
  Flat file serviceoperations Read (fileId, I, n) -> data Write (fileId, I, data) Create() -> fileId Delete (field) GetAttributes(FileId) -> attr SetAttributes(FileId, attr)
• 12.
  Terminologies  Access control Hierarchic file system  File group