Research And Implemention Of Automatic Expansion Clouds Storage System Based On ISCSI

Cloud storage is a hot direction of development of cloud computing, along with the development of Internet.With the amount of data on the Internet becaming more and more bigging, cloud storage is more and more close to the majority of Internet users to life, closely linked with our life and work, we often putside important data on Cloud SkyDrive prevent accidental loss. So most of the site provide basic cloud storage service,now. user obtain fixed some storage space after Registration, but with the increasing of user’s data, the storage space of the original can no longer meet the needs of users.The research of several current mainstream cloud storage mechanism within the industry, to propose a set of automatic expansion of cloud storage system based on iSCSI, providing a dynamic expansion strategy from client to server.This thesis studies some key technologies of designed principle of private cloud storage system, such as iSCSI, Samba, virtual logical volume, Quota, built a set to facilitate the implementation of theautomatic expansion of private cloud storage system based on these open source technology. The communication between server nodes, the analyse the concurrent models in common use, make full use of the more efficient mode l epoll and thread pool concurrency, besides compared with the traditional multi process.This system uses the remote storage to mount to local space, abandoned the space concept of the distance. The system adopts the three layer architecture.The bottom is the infrastructure layer, providing storage service. The management node locates this layer, namely providing storage service uniform for the upper node layer.At the upper layer is a data node layer. Data layer is the upper layer user nodes provide data access service.Data node layer occupied an important role in the system, which Samba file sharing service, quota user space quota with Lvm virtual logical volume, when a data node capacity utilization reaches a certain amount, the data nodes and management nodes communication,merging mount management node provides storage space and local, realizing the online capacity expansion objective.In order to make full use of system resource utilization, to propose expansion algorithm for dynamic monitoring and scanning, according to the user can write time number dynamically determine the next scanning time, furthermore this thesis uses the epoll and thread pool concurrency model. Finally, the article has carried on the system’s function and performance-test, proved the feasibility of the system.