Sie sind auf Seite 1von 1

ASDF is a distributed file system based on grid computing and peer to peer network.

The advantage of integrating grid and P2P network is to provide users with not only with large amount of storage space, but also an autonomous and scalable file system. Google File System and Hadoop File System are clientserver architecture; they consist of single metadata server that stores the metadata of all the files in the system and a number of data servers. Thereby, a single point of failure is there and bottlenecks may be caused on the metadata server. Scalability, reliability and availability are achieved by P2P architecture and to enhance extensibility and autonomy service oriented architecture is used. The services provided by the system are run as separate user level processes so that isolation, flexibility and extensibility can be provided. The four services provided by the system are: i. ii. iii. iv. Metadata Service: Metadata stores the attributes of the directories, files and symbolic links stored on the distributed file system. Metadata is cached locally in memory for future access. Data Service: It maintains file content and is responsible for contacting the local file system. Account Service: It is responsible for user authentication. Group Service: A mechanism for grouping the users who are working on the same projects thus sharing files and documents.

For communication FreePastry routing protocol is used, it is responsible for routing messages and locating services. Two types of messages are used; InternalMessage for local communication, ServiceMessage is exchanged outside local machine, it encapsulates InternalMessages and additional information like type and identifier of service. A client server model is used to define the relationship between metadata and data service. Each metadata service will maintain a list of data services. The metadata service serves as a monitor to inspect the status of the data services belonging to it. When the user puts a file on the system, the metadata service can query the status of the data service at that moment, and select suitable data services for storing data segments according to our systems policies.When a file is stored on ASDF, it is sliced into several data segments and these data segments are distributed over different data services. The data segments are also replicated which can be modified by a configuration file. For the privacy of information, the concept of super peer is used. Only special highly reliable, secure and trustworthy machines can become super peer and store account and group information. The proposed system provides concurrent transfer of multiple data segments by several data services, thus giving a performance increase in getting bigger files. Thus, it is highly suitable for accessing and storing massive amounts of data online.

Das könnte Ihnen auch gefallen