Title :
Object replication and migration policy based on OSS
Author :
Zeng, Ling-Fang ; Feng, Dan ; Wang, Fang ; Zhou, Ke
Author_Institution :
Key Lab. of Data Storage Syst., Huazhong Univ. of Sci. & Technol., Wuhan, China
Abstract :
It obviously gets limited advantage in massive storage system that all the management works are done by the servers, such as file server and metadata server of traditional storage area network (SAN). Moreover, storage device with deficient intelligence leads to lack of supporting local policy and does not scale well in performance and functionality when the number of storage device is added. At the same time, servers, e.g. metadata servers, are easy turned into bottlenecks when load balancing policies only depend on servers to scheme. This paper studied object storage system (OSS) which is a novel solution for massive storage. And the whole object storage system is divided into some domains. Each domain has its own metadata server, object storage controllers and clients (or application servers). Load balancing policies, involved three facets: metadata server (based on globe), object storage controller (based on local) and object, are presented.
Keywords :
client-server systems; file servers; resource allocation; storage area networks; storage management; file server; load balancing; massive storage; metadata server; object migration; object replication; object storage controller; object storage system; storage area network; storage management; Computer networks; File servers; Intelligent networks; Laboratories; Load management; Network servers; Peer to peer computing; Proposals; Scalability; Storage area networks; Load balancing; object migration; object replication; object storage system;
Conference_Titel :
Machine Learning and Cybernetics, 2005. Proceedings of 2005 International Conference on
Conference_Location :
Guangzhou, China
Print_ISBN :
0-7803-9091-1
DOI :
10.1109/ICMLC.2005.1526917