Key Technologies For Data Protection In Cloud Computing Environments

The rise of cloud computing brings both new challenges and new opportunities todata protection technologies.Cloud storage systems have large scale and hold huge data. They require morepowerful data protection technologies since their system complexity and operationcharacteristicsgreatly reduce the data reliability. Meanwhile, Cloud hosts face serioussafety threats of intrusions by hackers around the world. They require new dataprotection technologies since normal technologies cannot resist hackers’ sabotage.On the other hand, as a new kind of IT infrastructure resources, cloud resourceshave the advantages of high cost-effectiveness and high scalability, which make itpossible for existing data protection technologies to achieve better operation metrics.Therefore, this thesis focuses on and studies several key technologies for dataprotection in cloud computing environments. Main contents and major contributions ofthis thesis are summarized as follows:For the protection of data in cloud storage systems with deduplication, this thesisproposes a data reliability guarantee mechanism, which employs different algorithmsand schemes of erasure code to encode and protect deduplicated chunks withdifferentimportance. Experimental results show that it can achieve higher data reliability,lower storage cost and faster data recovery than traditional methods.For the protection of data in cloud hosts, this thesis presents a virtual disk basedintrusion detection mechanism, which reconstructsfile level metadata and operationinformation then detects intrusions within the virtualization platform under cloud hosts.Experimentsdemonstrate that it has good capability to detect intrusions while havingvery little effect on cloud hosts’ overall performance.For the block-level continuous data protection, this thesis proposes a novelapproach that stores version data into cloud. It adopts the data deduplication mechanism,a specific block version index structure, a disk/memory hybrid local cache and aquery-and-fetch recovery process, and can obtain both lower total cost and faster datarecovery than other approaches.For the data disaster recovery, this thesis presents a practical multi-cloud based data disaster recovery service model to overcome the failure of a single cloud. Thismodel utilizes resources of multiple clouds cooperatively and exposes a simple andunified service interface to hide the heterogeneity of clouds from users. And it cansimultaneously optimize data reliability, total cost of ownership and data recoveryspeed.In the end, this thesis introduces the design and implementation of an informationlifecycle management system, which uses various kinds of storage systems includingcloud storage systems to build a multi-levelstorage architecture. By using this system,information data can be stored, managed and protected hierarchically.