Research On Key Technologies Of IP-SAN Based Mass Data Storage

With the enterprise information increasing explosively, there are higher and higherdemands for storage capacity and reliability. It is critical for enterprises to have asolution to network storage. IP-SAN becomes one of the most important networkstorage technologies with its cheap cost and high performance.The research task of this dissertation comes from two actually developingprograms. One is an international cooperation program "Research on SAN-based massdata storage", The basic target of this program is to implement backup and managingmass data on cross-platform system. The other is a native cooperation program"SAN-based heterogeneous databases integrating system", the main purpose of thisprogram is to solve problems about Data Island, avoid/reduce data repetition, andrealize multidimensional use on unique data source.This dissertation is a conclusion of the author's research achievements in these twoprograms. The research work presented in this dissertation focused on solving mass datastorage with IP-SAN, including backup and disaster recovery in mass data storage, thesecurity of IP-SAN system, and the block scheduling algorithm about resourcemanagement in SAN-based system, mass data's separation and problems in itsheterogeneous data integrating. Then we implemented two application systems based onIP-SAN. The innovation of this dissertation including the following contents:1. SAN system structure based on cluster technology was proposedBased on the analysis of the development of SAN backup application, SAN systemstructure based on cluster technology was proposed, implemented two layers transparentfunction: first, user can access device transparently; second, the realization of scheduleris transparent to user. Not only realized the whole SAN system storage devicesvirtualization, but also realized optimizing the whole SAN system performance by theentire resource schedule of SAN system and the actual situation-of storage tasks. At thesame time, described the steps of accessing the storage devices, gave the advantage ofthe cluster technology-based SAN system structure.2. Presented a new scheduling algorithm based on quasi ordering for relatedbackup taskIn related backup task model, the end time of all the backup tasks has to bedelayed for the events of some waiting tasks haven't come. To solve this problem, wedefined a quasi ordering-based related backup task modol; implemented correspondingschedule algorithm.This backup model is mathematically validated to comply withquasi ordering. Furthermore, we observed from experiments that this algorithmimproves the efficiency and executing speed for storage backup tasks.3. A new secure IP-SAN system based on information hiding technology issuggestedFirst we gave a description of the conventional IP-SAN system structure, analyzedits security problems and crisis, on the basis of this, we implemented a new secureIP-SAN system model which based on information hiding technology. We described themain point of information hiding technology, showed its advantage by comparing withlocking technology. And then discussed algorithm about information hiding andretrieval in detail, imported information entropy as measuring standard in algorithmevaluating, the result shows that this new secure IP-SAN system has higher secure performance than conventional ones. Then, some methods about increasing the securityof IP-SAN system were discussed here.4. A SAN storage resource management model based on catalogue technology wasproposedOn the basis of analying SAN system model, we suggested a cataloguetechnology-based SAN resource management model. Described its data access methodin detail and analyzed the node performance of SAN system, researched on node offloadmeasuring standard, use fault prediction technology to dope out node trouble.5. A layered data description method was suggestedThe complexity of data description can be reduced by layered structure. Wesuggested a layer-based data description method, constructed mathematic model todata's physics-layer, structure-layer and semantic-layer; proved they were associatedwith each other. Based on this, we gave multi layer use to unique data source.Meanwhile, we implemented to use MDU in the plane design digitaled project bydeeply researched on document description of database system.6. Presented a XML-based algorithm of heterogeneous database conversionTo solve the data conversion problem in heterogeneous databases integrating, wedeeply researched the heterogeneous databases' feature and XML protocol, presented aXML-based heterogeneous database conversion strategy, especially described theprocess of heterogeneous database conversion based on XML, and showed thealgorithm.7. Implemented using layered heterogeneous data integrating algorithm whichbased on IP-SAN in digitized aircraft designFor the advantages of IP-SAN, we provided a heterogeneous data integratingalgrithem which took IP-SAN as data center and based on layered meaning,implemented the heterogeneous data integrating system in digitized aircraft designengineer, on the basis of showing the digram of the system, we described the systemintegrating layers and characteristic.The task of this dissertation is to research about IP-SAN system-based mass datastorage. The research achievements can be used in cosmically commercial application,scientific computing, network computing, high performance computing, aviationexploration, and data integrating and so on. Supply support for storage system design inthese Data-intensive fields. Therefore, the research of this dissertation is provided withsignificant theory meaning and engineering practical value.