preview

Nt1330 Unit 3 Assignment 1

Better Essays

ECE 677 Homework 5 Name : Ting-Chun Chiu ID : 23359631
1. Server state :
Server state can be divided into two part : the stateful and stateless server. The stateful server is that when a client open a file the server gives that client an unique identifier and stores client’s information into its memory. Although this method can improve performance, however, stateful server is generally be avoided in distributed system. On the other hand, stateless server uses totally different mechanism that the server identifies the file information and client position in each request but save nothing into its memory. The advantage is that it is easier to use fault tolerance on stateless server.
Granularity :
In distributed file system, there are two …show more content…

Comparison between NFS, AFS, GFS and HDFS : Network File System Andrew File System Google File System Hadoop Distributed File System
Server state Stateless server Stateless server Stateless server Stateless server
Granularity 8k chunk Whole file caching on client disk 64MB chunk 64MB block
Relationship Remote file access Upload/download service Remote file access Remote file access
Memory/disk usage Server side for disk access, client side for I nodes and file data. Client requests file from the cold cache of the server and stores file on its disk. Files are stored at server disk. Clients do read and write operations on the network. Clients contact HDFS through the NameNodes and fetch the file from DataNodes (server storage).
Consistency NFS doesn't provide concurrency-control mechanisms. AFS uses weak consistency model, which called call-back mechanisms. When write operation are done on existing file, the data is appended to end of the file. This method serializes the data write in GFS. HDFS uses NameNode operation to realize data consistency. NameNodes utilizes a transactional log file to record all the changes of

Get Access