Distributed File System Application (upper) theory

Since June, has been thinking about a set of common service platform. In the design of user behavior analysis and user promotion, found in their own framework for mass file storage does not have a reasonable solution. Originally intended to use Windows2003 DFS system to develop a new file system, and later found that the DFS under win is a big pit, attempted. Then consider the win platform and Linux system between the advantages and disadvantages of file processing and stability, the final choice of Linux under Fastdfs.

The following is a brief introduction to the Distributed File system and then combined with my actual case to give you pictures and texts, before the first thanks to the next Fishman, thump, and the strong support of students. You are the best!!

Tracker server: Tracking servers, mainly to do scheduling work, in the access load balanced role. Recording state information for group and storage servers in the cluster in memory is a hub for connecting client and storage servers.

Storage Server: Storage server, file and file attributes (meta data) are saved to the storage server

Architectural Interpretation:

Only two roles, tracker server and storage server, do not need to store file index information

All servers are equivalent, there is no master-slave relationship

Storage servers are grouped by the same files on the storage servers in the group (RAID 1)

Different groups of storage servers do not communicate with each other

Active reporting of state information by storage server to tracker server, tracker server does not normally communicate with each other

Upload file Flowchart:

1.client asked tracker uploaded to the storage;

2.tracker returns a usable storage;

3.client Direct and storage communication complete file upload, storage return file ID